Software Agents for Market Design and Analysis Project Team
- 格式:pdf
- 大小:953.20 KB
- 文档页数:82
LifeBook® T5010 Tablet PCThe comfort and ease of a traditional notebook combined with the versatility of a tablet.With its beautiful 13.3-inch widescreen display, built-in modular bay, integrated wireless access, and the latest Intel® Mobile Technology,you can quickly enter and access information whenever, wherever,and by whichever method you prefer.BENEFITSProduct VersatilityThe LifeBook T5010 Tablet PC is not only perfect for project-based users who require the speed and ease of forms-based input and pen-driven navigation, but also for mobile professionals who wantto maximize their productivity back at the office. With the optional dual digitizer, you can use your stylus or the lightest touch from your fingers to navigate around your display. This makes computing very intuitive, so sharing ideas is faster and easier. By using the Genuine Windows 7 operating system, you will be able to exploit the benefits of dual digitizer functionality with Touch Launcher. This Fujitsu exclusive feature can be customized to launch whatever youwant by simply touching an icon on your screen.Taking advantage of the industry-first bi-directional swivel hinge, you can easily transform this standard notebook into a tablet by rotating (in either direction) and folding the screen down. Collect data efficiently and accurately while in tablet mode. Use the powerful inking capabilities to take notes or capture a signature, then convert your handwritten notes into text. When you return to your hotel or office, use the keyboard in standard notebook mode to incorporate your notes into a business proposal or summary report. Maximum ProductivityWork without straining your eyes or your arms. The LifeBook T5010 Tablet PC weighs the same as many of its 12" competitors but features a big and bright 13.3" widescreen display that’s more than 20 percent bigger than a typical 12" widescreen display. The pen provides precise cursor control in any software application, including applications developed for special uses. Users can also mark up digital documents directly, annotate over images, conduct interactive presentations, sign forms, or even take handwritten notes. With Intel Core 2 Duo Technology, and the full power and enhanced functionality of Genuine Windows Vista® Business, the LifeBook T5010 Tablet PC delivers mobility, convenience, and increased productivity. Boundless ConnectivityConnect to your office network or the Internet with (optional) modem, Gigabit Ethernet, the latest Intel Wireless WiFi Link 5300AGN (802.11.a/b/g/draft-n) for up to 5X throughput improvement, or the optional Atheros® XSPAN wireless LAN. With the integrated wireless high-speed Broadband option, you can stay connected virtually anywhere to access your e-mail and office network with ease. If you use a Bluetooth-enabled cellular phone or PDA, your LifeBook T5010 Tablet PC can be equipped with integrated Bluetooth (version 2.1) wireless technology for data synchronization and swapping images.No-Compromise ComputingUse your LifeBook T5010 Tablet PC in notebook mode or in tablet mode, as your needs dictate. The built-in modular bay allows you to have a second battery for full work days and plane trips without adding much weight or increasing bulk. It can also accommodate one of the modular optical drives to handle all of your digital media needs: DVD movies, CD audio, or burning data onto CDs or DVDs. While in tablet mode, the exclusive patent pending BayLock™ utility automatically locks the media drive door closed and stops the drive from spinning, protecting both the drive and the media inside from accidental ejection or potential damage. This is a key feature for the mobile tablet PC user. All of this functionality comes in a tablet PC that weighs as little as 4.5 lbs., and still manages to house an array of ports that allow you to connect peripherals directly. An optional port replicator with hot-docking capability gives you immediate access to desktop peripherals. ManageabilityAvailable on select configurations, Intel Active Management Technology (AMT) is a feature of the Intel® Core™2 processor with vPro™ technology. It allows IT administrators to better safeguard their business from data loss and disruption. Using the hardware-based technology and an asset management solution, IT administrators can discover assets even while PCs are powered-off, reduce PC downtime by remotely isolating and recovering systems, proactively detect and verify if software agents are running and/or missing, detect and block incoming threats before they impact the network, and help keep software and virus protection up to date.SecurityThe LifeBook T5010 Tablet PC provides excellent security features which include: the LifeBook Application/Security Panel, an integrated fingerprint sensor, a security panel, embedded Trusted Platform Module (TPM) 1.2, a dedicated Smart Card slot, BIOS and HDD password protection, and embedded BIOS support for Computrace theft-protection software, and two security lock slots. These security systems are designed to be easy for you use, yet next to impossible to breach.Quality and ReliabilityThe close integration of Fujitsu design, manufacturing, and service creates some of the highest quality products offered in the PC market. In a time when most PCs are assembled by third-party vendors, Fujitsu takes pride in its own world-class design and manufacturing capabilities.OVERVIEWDual functionality for notebook and tablet useCrisp 13.3-inch Widescreendisplay illuminated byenergy-efficient LEDsBi-directional LCD hingefor everyday usabilityBuilt-in modular bay supports optical drives or bay battery for extended productivityIntegrated Wi-Fi Certifiedwireless LAN, Gigabit Ethernet, optional Bluetoothwireless and optional modem (not available onall configurations)Bluetooth wireless and integrated mobile wirelessBroadband (WWAN) areoffered as optionsDual digitizer display supports both pen andmulti-touch screen1Now available with Genuine Windows® 7 ProfessionalLAMP(S) INSIDE THIS PRODUCT。
26个英文字母开头的职业单词A. Accountant 会计师Accountants are professionals who are responsible for managing financial accounts for organizations, which include preparing financial statements, analyzing financial data, and advising on tax matters. They must possess strong analytical skills, attention to detail and a thorough knowledge of accounting principles and practices.B. Biologist 生物学家A biologist is a scientist who studies living organisms, their behavior, and characteristics. They may specialize in a particular area, such as marine biology, wildlife biology, or genetics. Biologists use a variety of tools and techniques to study organisms at different levels, from molecular biology to ecosystems.C. Chef 主厨Chefs are culinary professionals who manage kitchen operations and oversee food preparation. They may work in restaurants, hotels, or other food service establishments. Chefs must have excellent cooking skills and knowledge of food preparation techniques, as well as strong leadership and interpersonal skills to manage a team of kitchen staff.D. Dentist 牙医Dentists are medical professionals who specialize in oral health, treating and preventing oral diseases and defects. They perform a range of procedures, such as fillings, extractions, and root canals, and may also specialize incosmetic dentistry. Dentists must have strong communication and interpersonal skills, as they often work closely with patients to help them achieve optimal oral health.E. Engineer 工程师Engineers are professionals who design and build structures, machines, and systems that meet specific technological needs. They work in a variety of industries, including aerospace, automotive, biomedical, and civil engineering. Engineers must have a strong foundation in math and science, as well as the ability to apply that knowledge to solve complex problems.F. Farmer 农民Farmers are individuals who work in agriculture, producing crops and raising livestock. They may own their own farms or work for larger agricultural organizations. Farmers must possess knowledge of crop and animal husbandry techniques, as well as strong physical abilities to handle the demands of daily farm work.G. Graphic designer 平面设计师Graphic designers are professionals who create visual concepts or designs to communicate ideas to a target audience. They develop designs for a variety of media, such as advertisements, magazines, websites, and packaging. Graphic designers must have strong creative abilities, as well as proficiency with design software and a thorough understanding of design principles.H. Hair stylist 美发师Hair stylists are professionals who specialize in hair care and styling, working in salons or as freelance artists. They perform a range of services, including haircuts, coloring, and styling. Hair stylists must have strong creative abilities and knowledge of various hair care techniques, as well as good communication and interpersonal skills.I. Illustrator 插画师Illustrators create visual art that tells stories or communicates ideas, often for use in books, magazines, or other types of media. They may specialize in different areas, such as children's book illustration or medical illustration. Illustrators must have strong creative abilities, as well as proficiency with various artistic tools and techniques.J. Journalist 新闻记者Journalists are professionals who gather, investigate, and report news and current events. They may specialize in different areas, such as politics, sports, or business. Journalists must have strong research, writing, and communication skills, as well as the ability to work under tight deadlines and pressure.K. Kindergarten teacher 幼儿园教师Kindergarten teachers work with young children, usually between the ages of three and six, to promote learning and introduce them to basic concepts in reading, writing, math, and social skills. They must have strong communication and interpersonal skills, as well as a thorough understanding of how young children learn and develop.L. Lawyer 律师Lawyers are legal professionals who represent clients in various legal matters. They may specialize in different areas, such as corporate law or criminal law. Lawyers must have strong analytical and critical thinking skills, as well as excellent communication and interpersonal skills, in order to effectively represent their clients.M. Musician 音乐家Musicians are artists who create and perform music, whether as solo acts or in groups. They may specialize in different genres or types of music, such as classical, rock, or jazz. Musicians must possess strong creative abilities and technical skills, as well as the ability to work well with others in a collaborative setting.N. Nurse 护士Nurses are medical professionals who assist in the care of patients, working alongside physicians and other medical staff. They provide a range of services, such as administering medication, monitoring patient vital signs, and educating patients on basic health care. Nurses must have strong communication and interpersonal skills, as well as a thorough knowledge of medical procedures and terminology.O. Occupational therapist 职业治疗师Occupational therapists are medical professionals who help individuals improve their ability to perform daily tasks, such as eating, dressing, andbathing, due to physical, mental, or developmental disabilities. They work with patients to develop individualized treatment plans and may specialize in different areas, such as pediatrics or geriatrics. Occupational therapists must possess strong communication and interpersonal skills, as well as a thorough understanding of anatomy and physiology.P. Photographer 摄影师Photographers are artists who use cameras and lighting to create images that communicate a particular message or emotion. They may specialize in different areas, such as fashion photography or documentary photography. Photographers must possess strong technical skills, creativity, and attention to detail, as well as the ability to work well with others in a collaborative setting.Q. Quality control inspector 质量控制检查员Quality control inspectors are professionals who ensure that products meet specific quality standards before they are released to consumers. They may work in manufacturing, food production, or other industries. Quality control inspectors must have strong attention to detail and analytical skills, as well as a thorough understanding of quality control procedures.R. Real estate agent 房地产经纪人Real estate agents are professionals who assist clients in buying, selling, or renting properties. They must have strong communication and interpersonal skills, as well as a thorough understanding of the real estate market and legal procedures related to buying and selling property.S. Software developer 软件开发人员Software developers are professionals who design, develop, and test software applications for different platforms, such as desktop and mobile devices. They may work in various industries, such as gaming or finance. Software developers must have strong analytical and problem-solving skills, as well as proficiency in programming languages and software development tools.T. Teacher 教师Teachers are professionals who assist in the education and development of children and young adults. They may specialize in different areas, such as elementary education, secondary education, or special education. Teachers must have strong communication and interpersonal skills, as well as a thorough understanding of their subject matter and teaching methods.U. Ultrasound technician 超声技师Ultrasound technicians are medical professionals who use ultrasound technology to produce images of the inside of a patient's body. They may work in different medical settings, such as hospitals or clinics. Ultrasound technicians must have strong technical skills, as well as a thorough understanding of anatomy and physiology.V. Veterinarian 兽医Veterinarians are medical professionals who specialize in animal health and well-being. They may treat a variety of animals, such as domestic pets,livestock, and exotic animals. Veterinarians must have a strong understanding of animal anatomy and physiology, as well as excellent communication and interpersonal skills.W. Web developer 网页开发人员Web developers are professionals who design and develop websites, ensuring that they meet specific functionality and design requirements. They may work in various industries, such as e-commerce or marketing. Web developers must have strong technical skills, as well as proficiency in programming languages and web development tools.X. X-ray technician 放射技师X-ray technicians are medical professionals who use X-ray technology to produce images of the inside of a patient's body. They may work in hospitals, clinics, or private imaging centers. X-ray technicians must have strong technical skills and a thorough understanding of radiation safety procedures.Y. Yoga instructor 瑜伽教练Yoga instructors are professionals who teach students how to practice yoga poses and breathing techniques. They may work in yoga studios or fitness centers. Yoga instructors must have strong physical abilities, as well as a thorough understanding of yoga principles and techniques.Z. Zoologist 动物学家Zoologists are scientists who study animals, their behavior, and characteristics. They may specialize in different areas, such as animalbehavior or wildlife conservation. Zoologists must have a strong foundation in biology and a thorough understanding of animal anatomy and physiology.。
常见职业英语表达Common Occupations.The world of work is vast and varied, with countless job titles and industries to choose from. While some occupations are more common than others, all play a vital role in the functioning of society. Here are some of the most common occupations in the United States, along with their brief descriptions:1. Administrative Assistant.Administrative assistants provide administrative and clerical support to executives, managers, and other professionals. They perform a wide range of tasks, including answering phones, scheduling appointments, managing email, and preparing presentations.2. Cashier.Cashiers work in retail stores, restaurants, and other businesses where they process customer payments. They typically operate cash registers, handle transactions, and provide customer service.3. Chef.Chefs are responsible for planning, preparing, and cooking meals in restaurants, hotels, and other foodservice establishments. They oversee kitchen staff, ensure food quality, and create new dishes.4. Cleaner.Cleaners maintain the cleanliness of buildings, offices, and other spaces. They typically sweep, mop, dust, andempty trash cans.5. Customer Service Representative.Customer service representatives provide assistance to customers via phone, email, or chat. They answer questions,resolve complaints, and provide information about products or services.6. Data Entry Clerk.Data entry clerks input data into computer systems from various sources, such as paper documents, spreadsheets, or online forms. They ensure that data is accurate and complete.7. Delivery Driver.Delivery drivers transport goods and packages from one location to another. They may work for companies such as UPS, FedEx, or Amazon.8. Doctor.Doctors are medical professionals who diagnose andtreat illnesses and injuries. They may specialize in a particular area, such as family medicine, pediatrics, or surgery.9. Engineer.Engineers design, build, and maintain infrastructure, products, and systems. They may work in fields such ascivil engineering, electrical engineering, or mechanical engineering.10. Factory Worker.Factory workers operate machines and assemble products in manufacturing facilities. They may work in industries such as automotive, food processing, or electronics.11. Hair Stylist.Hair stylists cut, color, and style hair in salons and barbershops. They may also provide other services, such as hair extensions and scalp treatments.12. Human Resources Manager.Human resources managers oversee the recruitment, hiring, and development of employees. They also manage employee benefits, payroll, and compliance with labor laws.13. Information Technology (IT) Specialist.IT specialists provide technical support and maintenance for computer systems and networks. They may work in fields such as network administration, software development, or cybersecurity.14. Lawyer.Lawyers represent clients in legal matters. They may specialize in areas such as criminal law, civil law, or corporate law.15. Manager.Managers supervise and coordinate the work of other employees. They may be responsible for a specific department, project, or team.16. Marketing Manager.Marketing managers develop and execute marketing campaigns to promote products or services. They conduct market research, manage advertising budgets, and track results.17. Nurse.Nurses provide medical care to patients in hospitals, clinics, and other healthcare facilities. They may work in areas such as intensive care, pediatrics, or geriatrics.18. Police Officer.Police officers enforce laws, prevent crime, andprotect the public. They may work in various settings, such as patrol cars, on foot, or on bicycles.19. Real Estate Agent.Real estate agents help clients buy, sell, or rent properties. They provide guidance throughout thetransaction process and represent their clients' interests.20. Software Developer.Software developers design, develop, and maintain software applications. They may work in fields such as web development, mobile app development, or data science.21. Teacher.Teachers instruct students in various subjects at schools, colleges, and universities. They develop lesson plans, deliver lectures, and assess student performance.22. Truck Driver.Truck drivers transport goods and materials over long distances. They may work for companies such as trucking companies, freight forwarders, or delivery services.23. Waiter/Waitress.Waiters and waitresses serve food and drinks to customers in restaurants and other food service establishments. They take orders, deliver food, and provide customer service.These are just a few examples of the countless occupations that exist in today's job market. Each occupation offers its own unique set of responsibilities, challenges, and rewards. Whether you are just starting your career or looking for a change, exploring the different options available will help you find the perfect fit for your skills and interests.。
Chapter 11.Describe three factors that would cause a company to continue doing business intraditional ways and avoid electronic commerce.Traditional commerce is a better way to sell the items or services when personal selling skills are a factor, as in commercial real estate sales; or when the condition of the products is difficult to determine without making a personal inspection, as in the purchases of high-fashion clothing, antiques, or perishable food products.2.What are transaction costs and why are they important?Transaction costs are the total of all costs that a buyer and seller incur as they gather information and negotiate a purchase-sale transaction.6.How might a manager use SWOT analysis to identify new applications for electroniccommerce in their strategic business units?In SWOT analysis you list strengths and weaknesses of the business unit and then identify opportunities presented by the markets of the business unit and threats posed by the competitors of the business unit. This is accomplished by analyzing all the business operations into value adding activities and supporting activities.7.In about 200 words, explain the difference between language translation and languagelocalization.Language translation is the process of restating some text written in one language in a different language. In other words, to translate is examine some original text, written in what is called the source language, and to write a corresponding text in different language, called the target language, with the goal of preserving the tone and meaning of the original text.Language localization is a translation that considers multiple elements of the local environment, such as business and cultural practices, in addition to local dialect variations in the language. The cultural element is very important since it can affect—and sometimes completely change—the user’s interpretation of text.8.In a paragraph, describe the advantages of a flat-rate telecommunications access systemfor countries that want to encourage electronic commerce.In a flat-rate access system, the consumer or business pays one monthly fee for unlimited telephone line usage. Although many factors contributed to the rapid rise of U.S. electronic commerce, many industry analysts agree that flat-rate access has been one of the most important factors. As more European telecommunications providers have begun to offer flat-rate access, electronic commerce in those countries has increased dramatically.1.You have decided to buy a new laser printer for your home office. List specific activitiesthat you must undertake as you gather information about printer capabilities andfeatures. Use the CompUSA, , Office Depot, OfficeMax, and Staples Web sites to gather information. Write a short summary of the process you undertook so that others who plan to undertake a similar task can use your information.Answers will vary, but should include the following:∙Identify a search engine∙Visit Websites for information on features, attributes and benefits∙Interrogate Websites for additional information∙Place an order∙Inquire about shipping/payment terms2.Choose one of the Web sites listed in the previous question and identify three ways inwhich the company has reduced its transaction costs by using a Web site to provideinformation about printers. List these three transaction cost reduction elements and write a paragraph in which you discuss one transaction cost reduction opportunity that you believe the company missed.Answers will vary, but may include the following:∙Description of the item∙Employees no longer have to search for prices∙Delivery optionsStaples could include informational links such as: laser printer buying guide or laser printer maintenance.Chapter 21.What were the main forces that led to the commercialization of the Internet?Summarize your answer in about 100 words.The Internet was born out of the need for the U.S. government, specifically the Defense Department, to communicate with its weapons installation distributions all over the world.This idea, in the hands of researchers and scientists evolved even further allowing those researchers the capability to communicate with their colleagues at other universities. As personal computers became more popular and affordable, companies increasingly wanted to construct their own networks. This all led to the dramatic increase in business activity, but the commercialization of the Internet was really spurred by the emergence of the World Wide Web. The software that allowed computers to communicate while on the Internet is still the largest category of traffic today.2.Describe in two paragraphs the origins of HTML. Explain how markup tags workin HTML, and describe the role of at least one person involved with HTML’sdevelopment.SGML is a software language for describing electronic documents and how they should be formatted as well as displayed. This language is the precursor of HTML, which is used by all documents on the Web. Robert Calliau and Tim Berners-Lee independently invented HTML at the CERN research center in Switzerland. HTML’s document type definition is easier for users to learn and use for describing formatting and displaying electronic documents by Meta tag codes.3.In about 200 words, compare the POP e-mail protocol to the IMAP e-mail protocol.Describe situations in which you would prefer to use one protocol or the other andexplain the reasons for your preference.A POP message can tell the e-mail computer to send mail to the user’s computer and to dele teit from the e-mail computer; to send mail to the user’s computer and do not delete it; or simply to ask whether new mail has arrived. IMAP protocol performs the same basic functions as POP, but includes additional features that can instruct the e-mail server to send only selected messages to the client instead of all messages. It also allows the user to view only the headers and the e-mail sender’s name before deciding to download the entire message. One would choose IMAP if they have a need for a more robust system that allows them to access their email from different computers at different times.1.Bridgewater Engineering Company (BECO), a privately held machine shop, makesindustrial-quality, heavy-duty machinery for assembly lines in other factories. It sells its presses, grinders, and milling equipment using a few inside salespeople and telephones.This traditional approach worked well during the company’s start-up years, but BECO is getting a lot of competition from abroad. Because you worked for the company during the summers of your college years, BECO’s president, Tom Dalton, knows you andrealizes that you are Web savvy. He wants to form close relationships with the steelcompanies and small parts manufacturers that are BECO’s suppliers so that h e can tap into their ordering systems and request supplies when he needs them. Tom wants you to investigate how he can use the Internet to set up such electronic relationships. Use the Web and the links in the Online Companion to locate information about extranets and VPNs. Write a report that briefly describes how companies use extranets to link their systems with those of their suppliers, then write an evaluation of at least two companies (using information you have gathered in your Web searches) that could help develop an extranet that would work for Tom. Close the report with an overview of how BECO could use VPN technologies in this type of extranet. The three parts of your reportshould total about 700 words.Responses can vary significantly in this exercise. Any recommendation for systems development should include the infrastructure required to support a supply-chain management extranet, as well as the costs and the anticipated benefits. The infrastructure for a private network requires a TCP/IP network, Web authoring software, and a firewall server. The benefits include lower communication costs, and more timely and accessible information, as well as convenient use.2.Frieda Bannister is the IT manager for the State of Iowa’s Department of Transport ation(DOT). She is interested in finding ways to reduce the costs of operating the DOT’svehicle repair facilities. These facilities purchase replacement parts and repair supplies for all of the state’s cars, trucks, construction machinery, and road maint enanceequipment. Frieda has read about XML and thinks that it might help the DOT send orders to its many suppliers throughout the country more efficiently. Use the Online Companion links, the Web, and your library to conduct research on the use of XML in state, local, and federal government operations. Provide Frieda with a report of about 1000 words that includes sections that discuss what XML is and explain why XML shows promise for the ordering application Frieda envisions. Your report should also identify other DOT business processes or activities that might benefit from using XML. Thereport should also include a summary of the main disadvantages of using XML today for integrating business transactions. End the report with a brief summary of how the W3C Semantic Web project results might help the DOT operate more efficiently in the future.Responses should include the following points:▪XML uses markup tags to describe the meaning or semantics of the text.▪XML records are embedded in HTML documents.▪With XML, tags can be created that identify all the record details for the ordering application that Frieda envisions.▪The extensibility of XML is also its weakness. Sharing data across organizations means that the organizations must use the same tag na mes. For example, Frieda’s organizationmight create a tag called "PurchaseOrderNumber", and one of her suppliers might call the same item "OrderNum".▪The W3C Semantic Web will allow XML tags to be read by software agents, which will result in better, less-timely searching on the Web for information. This would allow theDOT to research pricing, availability, etc. of parts from suppliers.Chapter 31.Write a paragraph in which you describe the conditions under which a Web site canhope to become profitable if it relies exclusively on advertising revenue. In a secondparagraph, provide an example of a company not mentioned in the chapter that is using the advertising-supported model and that is likely to be successful in the long run.Explain why you think it will succeed.A Web site that relies exclusively on advertising revenues must contend with two majorproblems. First, no consensus has emerged on how to measure and charge for site views and the second problem is that very few Web sites have sufficient numbers of visitors to interest large advertisers. To be successful using the advertising-supported revenue model exclusively,a Web site must be large search engine, because they generate sufficient traffic to be profitable.Another alternative is to become a site that targets niche markets, for example employment-advertising sites.2.Describe two possible service-for-fee offerings that might become available tousers of Internet-enabled wireless devices (such as PDAs or mobile phones) in the near future. Write one paragraph for each service in which you outline the profit potential and risk of losses for each.Two possible fee-for-service offerings that might become available are medical and legal services. The profit potential for these services would be higher than the traditional brick and mortar services offered. There are limitations placed on these services because of licensure issues and dispensing legal and medical advice over the Internet is still a major hurdle. As technology and the Internet mature, it will be possible to offer these services offer the Internet.3.In two paragraphs, explain why a customer-centric Web site design is soimportant, yet is so difficult to accomplish.An important part of a successful electronic business operation is a Web site that meets the needs of potential customers. It is a significant challenge however, to design an effective website that introduces the company to different audiences (shareholders, the financial community, suppliers, potential alliance partners, potential customers, current customers and so on) with very different interests.4.Many real estate agents today have Web sites that list the properties they have for sale.These agents also advertise the properties in classified newspaper ads and sometimes in television ads. Write three paragraphs in which you briefly describe the things that realestate agents can best accomplish through (1) their Web sites, (2) mass media advertising, and (3) personal contact.Responses may be similar to the following:Web sites: Offer mortgage loan seekers online credit review and decisions within minutes.Mass media advertising: Offer property listings.Personal contact: Provide more detailed information about the property and about obtaining mortgages,1.Evaluate the usability of two Web sites that sell large-screen televisions. A list of links tocompanies that sell this product is included in the Online Companion for this exercise, but you may use other sites if you wish. In your evaluation, compare the sites on how easy it is to learn about the product and purchase the product. Your report shouldinclude a section of about 200 words in which you describe the criteria you used in your evaluation, a section of about 300 words that summarizes your findings, and a section of about 100 words in which you present your conclusion.Criteria that can be used to evaluate the Web sites include:∙Clarity of product information∙Prices relative to other online merchants∙Overall look and design∙Charges stated clearly before order submission∙Variety of shipping options∙Shipping chargesChapter 41.In about 600 words, explain the differences between customer acquisition and retentionand outline two marketing strategies that would help a company accomplish each ofthese two objectives. Be sure to present facts and logical arguments that support the use of each strategy for each objective.Customer retention is about making sure existing customers keep buying from you. On the Net, customer retention also means making sure your site visitors keep returning. Customer acquisition implies attracting new visitors to your Web site.2.Select a retail store with which you are familiar that has a Web site on which it sellsproducts or services similar to those it sells in its physical retail stores. Explore the Web site and examine it carefully for features that indicate the level of service it provides.Using your experience in the physical store and your review of the Web site, write a200-word evaluation of the company’s touchpoint consiste ncy.Responses will vary but a review should consider the following:The goal of providing the same quality of service is known as touchpoint consistency. The five levels are: awareness (customers who recognize the name of the company or its products), exploration (potential customers learning more about a company’s products and services), familiarity (customers who have completed several transactions and are aware of company policies), commitment (customers with preferences for the product; these customers are loyal and are willing to tell other potential customers), and separation (customers that are leaving, or separating, from the company for any reason).3.Many people have strong negative reactions to pop-up, pop-behind, interstitial, and richmedia ads. Write a 200-word letter to the editor of an Internet industry magazine inwhich you explain, from the advertiser’s viewpoint, why these ads can be effectiveadvertising media.Responses will vary, but the students might discuss how pop-behind ads remain visible after the browser has closed and keeps that information fresh in the user’s mind. In addition, a good point to bring up is that unlike pop-up ads, they do not cover the browser window.1.Visit the RedEnvelope Web site to examine how that company implements occasionsegmentation. Write a report of no more than 200 words in which you describe two clear examples of occasion segmentation on the site.Answers will vary, but Red Envelope has a section that displays various holidays, such as spring, Easter, and birthdays. Students should discuss the details of these different occasions.2.Marti Baron operates a small Web business, The Cannonball, that sells parts, repair kits,books, and accessories to hobbyists who restore antique model trains. Many model train hobbyists and collectors have created Web sites on which they share photos and other information about model trains. Marti is interested in creating an affiliate marketing program that would allow those hobbyists to place links on their sites to TheCannonball and be rewarded with commissions on sales that result from visitorsfollowing those links. Examine the services offered by Be Free, Commission Junction, LinkShare, and any other affiliate program brokers you can find on the Web.Recomme nd at least one affiliate program broker that would be a good fit for Marti’s business. In about 500 words, explain your recommendation. Be sure to consider thecharacteristics of Marti’s business in your analysis.The students should search different program brokers and try to find one that already deals in hobbies and collectibles. Be Free would be a good fit, since it caters to smaller businesses and is scalable as well.。
1.A user interface we said here is __ABC________A.a text-based user interface or GUIB.an interface between a computer and its peripheral deviceC.an interaction between an operating system and a userD.an interaction between an application program and a user2.___A___provides transparent transfer of data between end users, providing reliable data transfer services to the upper layers.A.The Transport LayerB. Session LayerC. Network LayerE.Application Layer E. Presentation Layer3.Many viruses do harmful things such as (ABCD ).A.deleting filesB. slowing your PC downB.simulating typos D. changing random data on your disk4.We can classify programming languages under two types:(AB ) languages and ( )languages.A.high-levelB. low-levelC. advanced-levelD. basic-level5.With an Internet connection you can get some of the basic services available are:___ABCD_______A.E-mailB. TelnetC. FTPD. Usenet news6. A general purpose computer has four main sections: ( ABCE).A.the control unitB. the memoryC. the input and output devicesD. the cpuE. the arithmetic and logic unit (ALU),7.Windows 2000 has the key technologies, they are (ABCD ).A.securityB. active directoryC. flat directoryD. enterprise management8.The register file is___ACD_______A.addressed by much shorter addressesB. physically largeC.physically smallD. one the same chip as the CPU9. A stack protocol can be used for (A ).A.removing the latest element ins( )ertedB. removing the earliest element ins( )ertedC. subroutine callsD. operation of arithmetic expressions10.The end equipment in a communication system includes (ABCD ).A.printersB. computersC. CRTsD. keyboards11.Microsoft Office Professional 2000 include____ABCD______.A.Excel 2000B. PowerPoint 2000C. Word 2000D. Outlook 200012. A general purpose computer has four main sections: ______ABCE______A.the input and output devicesB. the memoryC. the arithmetic and logic unit (ALU),D. the cpuE. the control unit13.The two most common types of scanners are (BC ) and ( )A. hand-held scannersB. flatbed scannersC. auto scannersD. handler scanners14.Some viruses use (CD ) and ( ) techniques to hide their existence.A.quickly spreadB. replace a part of system softwareC. stealthD. polymorphic15.The Windows 2000 product line includes____ABCD______.A.Windows 2000 Datacenter ServerB. Windows 2000 ProfessionalC. Windows 2000 ServerD. Windows 2000 Advanced Server16.Similar to viruses, you can also find malicious code in (ABC ).A.Trojan HorsesB. logic bombsC. wormsD. Microsoft Word Documents17.Viruses all have two phases to their execution, the ( ) and the ( BD).A.create phaseB. attack phaseC. del( )ete phaseD. infection phase18.Active Directory can help you (ACD ).A.get off the limits of down level networksB. deliver complete enterprise security by itselfC. build a complex international networkD. manage every resource with a single logon19.High-level languages are commonly classified as (ACDE ).A.object-orientedB. automaticC. functionalD. logic languagesE. procedure-oriented20.(CD )is a type of executable file .A.TXT fileB. JPG fileC. EXE fileD. COM file21.( ABCD) maybe a certain cause that some viruses infect upon.A.an external event on your PCB. a dayC. a counter within the virusD. a time22.(BC )is a type of executable file .A.TXT fileB. EXE fileC. COM fileD. JPG file23.The web browsers which is normal used, such as(ABCE ).A.FirefoxB. Internet ExplorerC. OpraD. ICQE. Apple Safari24.Newer ideas in computing such as(ABDE ) have radically altered the traditional concepts that once determined program form and functionA.artificial intelligenceB. distributed computingC. software engineeringD. parallel computingE. data mining25.Microsoft Windows currently supports __AC___and _____file systemsA.NTFSB. OCFSC. FATD. ext2E. NILFS26.Modem is ____ACD______.A.a modulator/demodulatorB. a data setC. a demodulatorD. a modulator27.The equipment _AB_____.A.transfers the number of bits in serial formB.manipulates digital information internally in word unitsC.transfers the number of bits in parallelD. manipulates digital information internally in serial form28.Electronic commerce that is conducted between businesses is referred to as business-to-business or DA. C2CB. C2BC. e-commerceD. B2B29.The World Wide Web also subsumes previous Internet information systems such as (AC ).A.GopherB. FtpC. FTPD. Telnet relies on the services of .NET data providers.There are ABCDA.ConnectionB. Data AdapterC. DataReaderD. Command31.The development process in the software life cycle involves four phases: analysis, design, implementation, and ___ACDE_____.A.analysisB. auditC. implementationD. designE. testing32.The end equipment in a communication system includes __ABCD____.A.printersB. CRTsC. computersD. keyboards33.In electronic commerce ,information search and discovery services include (ABCDE ).A.search enginesB. information filtersC. software agentsD. directoriesE. electronic catalogs34.GIS work with two fundamentally different types of geographic models.They are the (BD ).A.geography modelB. vector modelC. mathematic modelD. raster modelE. data model35.The two most common types of scanners are ____AC___and _____A.flatbed scannersB. hand-held scannersC. auto scannersD. handler scanners36.Windows 2000 has the key technologies, they are ( ABCD).A.active directoryB. flat directoryC. enterprise managementD. securityputer software, or just software is a general term used to describe a collection of computer programs, procedures and documentation that perform some tasks on a computer system.The term includes: ABCA.Firmware which is software programmed resident to electrically programmable memorydevices on board mainboards or other types of integrated hardware carriers.B.Application software such as word processors which perform productive tasks for users.C.Middleware which controls and co-ordinates distributed systems.38.Software engineering is related to the disciplines of ___ADE_____A.project managementB. natural languageC. netural networkD. systems engineeringE. computer science39.What make it is difficult to agree on how to count viruses? ABCDA.some viruses can create different versions when they infect other programsB.just a trivial change may create a new virusC.some viruses can use polymorphic techniqueD.new virus arise from an existing virus40. A virus is a program that reproduces its own code by (ABC ).A.ins_erting into the middle of a fileB. simply placing a pointerC. adding to the end of a fileD. replacing another program41.Input devices include: ____ABCD______.A.the keyboardB. microphoneC. touch screenD. the mouse42.Viruses all have two phases to their execution, the ( ) and the ( AD).A.attack phaseB. create phaseC. del_ete phaseD. infection phase43.The equipment (BC ).A.manipulates digital information internally in serial formB.manipulates digital information internally in word unitsC.transfers the number of bits in serial formD.transfers the number of bits in parallel44.Office automation is___CD_______.A.. the computerB. communications technologyC. the application of computerD. used to improve the productivity of people45.The types (classes, structs, enums, and so on) associated with each .NET data provider are located in their own namespaces are: ABCDEA.System.Data.SqlClient. Contains the SQL Server .NET Data Provider types.B.System.Data.Odbc. Contains the ODBC .NET Data Provider types.C.System.Data. Contains provider-independent types such as the DataSet and DataTable.D.System.Data.OracleClient. Contains the Oracle .NET Data ProviderE.System.Data.OleDb. Contains the OLE DB .NET Data Provider types.46.C++ is __ACD________.A.extended from CB. a superset of CC. object-orientedD. procedure-oriented47.Some viruses, with no attack phase, often damage the programs or disks they infect because they (AD ).A.have bugs in themB. show messages on your screenC. steal storageD. contain poor quality code48.Windows 2000 is ( A).A.an inventive technologyB. used for building a scalable networkB.the same as Windows NT D. new lease of Windows49.Some common applications related to electronic commerce are the following:A.TeleconferencingB. Online bankingC. EmailD. Instant messagingE. Enterprise content managementF. NewsgroupsG. Shopping cart software50. A program is a sequence of ( ) that can be executed by a computer.It can either be built into the hardware or exist independently in the form of (BC ).A.hardwareB. softwareC. instructionsD. data51.Electronic payments include ___ABCD______.A.credit card paymentsB. electronic checksC. digital currenciesD. cash payment52.The web browsers which is normal used, such as____ABCD_____.A.OpraB. Internet ExplorerC. FirefoxD. Apple SafariE. ICQputer programs fall into two major classes: __AC____and ____.A.application programsB. application suiteC. operating systemsD. database application54.Database connection ( B) allows an application to reuse an existing connection from a pool instead of repeatedly establishing a new connection with the database.A.pondB. poolingC. linkD. connection55.The development process in the software life cycle involves four phases: analysis, design, implementation, and (ABCE ).A.implementationB. designC. analysisD. auditE. testing56.Hypermedia include ( ABCDEF)A.video clipsB. imagesC. textD. flashE. vidoeF. sounds57.An asleep state is_____ABD_____ed to lessen wear-and tear on the computerB. used for saving energyC. indicated by the indicator light putting outD. a low power standby mode58.Electronic payments include (ABCD ).A.digital currenciesB. electronic checksC. credit card paymentsD. cash payment59.You want to (BCD ) maybe wake up a virus that has residented in memory.A.del( )ete a fileB. access a disketteC. execute a programD. copy a file60.Before you turn the power on with a new computer, you should make sure_____ABCD_____A.the computer system has been set upB. the computer is already out of the boxB.appropriate software have been installed D. appropriate cables are correctly connected61.Security is usually enforced through ___ABE________.A.access controlB. encryptionC. data retrievingD. data storingE. auditingputer programming is the process of ABCD__the source code of computer programs.A. testingB. maintainingC. debuggingD. writing63.Queues that occur everyday life can be seen ( ABCD).A.as automobiles waiting for a traffic lightB. as people waiting for service at a bankC. in certain societies lacking equalityD. in an emergency room of a hospital64.Static graphics include____AB______.A.animatorsB. photographsC. moviesD. pictures66.which is the threat to computer security in the following choices ? ABCDA. Computer CriminalsB. Human ErrorsC. Computer CrimeD. earthquake65.The attributes of the stack are______A____.A.queueB. FIFOC. built into their circuitryD. LIFO66.If a virus simply reproduce and have no cause for an attack phase, but it will still ( ) without your permission. BDA.play musicB. stealing storageC. del( )ete filesD. pilfer CPU cycles67.According to the text,modern digital computers can be divided into four major categories on the basis of cost and performance.They are _______ABDE___________. A.minicomputers B. mainframes C. note book D. workstation E. microcomputers F. lenovo68.The Application layer in TCP/IP Model is correspond to (ABD ) in OSI ModelA.Presentation LayerB. Session LayerC. Transport LayerD. Application LayerE. Network Layer69. A computer system user generally more cares for___ABD_______A.speed of computationB. storage sizeC. physical size of the computerD. efficiency of the computer71.Cache is____ABC______A. slowB. high costC. fastD. relatively small72.We can say a bus is simply_____ABC_____ .A. a wireB. a 16-bit busC. a group of wiresD. a 8-bit bus73.Viruses can delay their attack for (ABCD ).A.yearsB. monthsC. weeksD. days74.In order to increase our computer’s performance we need to____BCD______A.buy a L1 cacheB. have a much larger main memoryC. have a L2 cacheD. buy a L2 cache75.The software that controls the interaction between the input and output hardware is called BIOS,which stands of __A________A.Basic Input Output SystemB. Classic Input Output SystemB.Advanced Input Output System D. Junior Input Output System76.To enhance performance of a computer system we should____ACD______A.improve the pattern of referencing operandB. optimize the simple movement of dataC. optimize the basic sequence control mechanismD. use IF and LOOP instructions as many as possible77.Their company used international lawyers to prosecute a crime ring involving software ____A____in Thailand.A.piracyB. copyingC. duplicationD. cloning78.The software that controls the interaction between the input and output hardware is called BIOS,which stands of (B)A.Advanced Input Output SystemB. Basic Input Output SystemC. Classic Input Output SystemD. Junior Input Output System79.Some viruses use ( AC) and ( ) techniques to hide their existence.A.stealthB. quickly spreadC. polymorphicD. replace a part of system software80.Middleware lies in______ACD____A.the middle of interactions between different application programsB.the top of the layering vertical stackC.the top of an operating systemD.the middle of the layering vertical stack81.Software includes ( ACDE) etcA.video gamesB. all kinds of filesC. programsD. websitesE. mobile application82.The major functional components of an office automation system include: ___ABCD__A.electronic mailB. personal assistance featuresC. information storage and retrievalD. text processing83.The Internet carries various information resources and services, such as (ACDEF ) and the inter-linked hypertext documentsA.online chatB. talkingC. electronic mailD. file transferE. online gamingF. file sharing84. A processor is composed of:____ABCD______.A.an arithmeticB. a control unitC. RegistersD. logic unit85.Functions of the compiler used in RISC are ___ABC_______A.to optimize register usageB.to maximize register usageC.to allocate registers to those variables that will be used the most in a given time periodD.to compile a high level language program86. A digital computer is generally made up of five dstinct elements: a central processing unit,(ABCD).A.a busB. input devicesC. memeory storage devicesD. output devicesE. crt screen87.There are AB (CD)_between the DTEs.A.digital-to-analog converterB. the modemC. communications equipmentD. will be replaced by an upd_ated standard88.What make it is difficult to agree on how to count viruses? ABCDA.just a trivial change may create a new virusB.some viruses can use polymorphic techniqueC.some viruses can create different versions when they infect other programsD.new virus arise from an existing virus89.which aspect have to be considered in the design of a piece of software. ABCDEFGA.Fault-toleranceB. ExtensibilityC. ModularityD. CompatibilityE.MarketabilityF. PackagingG. Maintainability90.Active Directory can help you (ACD ).A.build a complex international networkB. deliver complete enterprise security by itselfC.manage every resource with a single logonD. get off the limits of down level networks91.Early computer solved_____CD_____ problems.A.controlB. engineeringC. mathematicalD. business applications92.The tools which Programming software usually provides include: ABCDEA.debuggersB. text editorsC. linkersD. compilersE. interpreters93.DTE is ( AB).A.data terminal equipmentB.the last piece of equipment that belonged to the subscriber in a data link systemC.satelliteD. Digital T-carrier94.According to the text,modern digital computers can be divided into four major categories on the basis of cost and performance.They are ( BDEF).A.note bookB. microcomputersC. lenovoD. minicomputersE. workstationF. mainframes95.which is the type of electronic commerce in the following choice ACA.B2BB. C2CC. B2C96.The operations of a structured data type might act on (ABCD ).A.a stackB. the values of the data typeC. component elements of the data structureD. a queue97.Types of media include__ACD________.A.textB. animationC. audioD. full-motion video98. A virus is a program that reproduces its own code by (ABC ).A.simply placing a pointerB. adding to the end of a fileC. ins( )erting into the middle of a fileD. replacing another program99.According to the text,the author mentions three of the most commonly used types of printer.They are (BDE ).A.belt printerB. dot-matrix printers;C. array printerD. laser printerE. inkjet printers100.The end equipment in a communication system includes ___ABD_______A.keyboardsB. DCEC. CRTsD. computers101.Software includes _____ACDE________etcA.programsB. all kinds of filesC. video gamesD. websitesE. mobile application102.With .NET, Microsoft is opening up a channel both to ( ) in other programming languages and to ( BC). (developers; components)A.coderB. developersC. componentsD. architecturemon contemporary operating systems include (ABCD ).A.LinuxB. Microsoft WindowsC. SolarisD. Mac OS104.A mechanism for translating Internet hostnames into IP addresses is___BCD_______A.equipped into the general-purpose operating systemB.typically inside of operating system kernelC.as a middleware by author’s definitionD.typically outside of operating system kernel105.RISC is____ABC______ed for many computer manufacturers nowadaysB.guided to be built from studying the execution behavior of high-level language programsC.abbreviation of reduced instruction set computerD.abbreviation of complex instruction set computer106.With .NET, Microsoft is opening up a channel both to _BC_______in other programming languages and to ________. (developers; components)A.coderB. componentsC. developersD. architecture107.The tools which Programming software usually provides include: ABCDEpilersB. interpretersC. text editorsD. linkersE. debuggers108.The following products of software are belong to middleware____BCD______A.OracleB. IBM’s Web Sphere MQC. Java 2 PlatformD. J2EE109.The system manager used by a fast processor can____BCD______A.connect a networkB. monitor processor’s core temperatureC. monitor processor’s supply voltagesD. reset a system110.Queues that occur everyday life can be seen (ABCD ).A.as automobiles waiting for a traffic lightB. as people waiting for service at a bankC. in an emergency room of a hospitalD. in certain societies lacking equality111.C++ include the following pillars: ____ABCD______.A.data hidingB. polymorphismC. encapsulationD. inheritance112.Windows 2000 is____ACD______A.new lease of WindowsB. an inventive technologyC. the same as Windows NTD. used for building a scalable network113.We use paged virtual memory to___ABCD_______A.extend the size of memoryB. reduce latency of the diskC. store large program and data setD. increase bandwidth of the disk114.According to the physical size of computers we can classify the __ABCD____ computers into A. supercomputer B. minicomputer C. microcomputer D. mainframe115.Some common applications related to electronic commerce are the following: ABCDEFGA.EmailB. TeleconferencingC. Instant messagingD. Shopping cart softwareE.NewsgroupsF. Enterprise content managementG. Online banking116.One machine cycle in RISC has _B_________A.two machine instructionsB. one machine instructionC. four machine instructionsD. three machine instructions117.The function of computer hardware is typically divided into three main categories.They are____ADE_____.A.inputB. motherboardC. cpuD. storageE. output118.Active Directory supports ( ABCD).A.granular access controlB. inheritanceC. encapsulationD. delegation of administrative task119.The core of SQL is formed by a command language that allows the (ACDE ) and performing management and administrative functions.A.deletion of dataB. process of dataC. updating of dataD. retrieval of dataE. ins( )ertion of data120.Some commentators say the outcome of the information revolution is likely to be as profound as the shift in (ABCD )A.industrialB. agriculturalC. Service IndustryD. handicraft industry。
A MOBILE AGENT-BASED COMMUNICATIONSMIDDLEWAREFOR DATA STREAMING IN THE BATTLEFIELDABSTRACTIn this paper we introduce the FlexFeed framework in the context of military combat operations. FlexFeed realizesthe notion of Agile Computing for streaming data communicationsand implements a flexible, robust and efficient]publish/subscribe infrastructure for dynamic ad hoc environmentunder resource and policy constraints. Theframework uses mobile software agents for underlyingconfiguration and policy enforcement. The paper illustratesthe effectiveness of the framework with quantitativeexperiments over simulated scenarios.INTRODUCTIONDependable communication capabilities are amongst the most important technical requirements for mission successin military operations. Complex missions involving coalitionforces, robotic support units, remote sensor beds andautonomous vehicles will require underlying communicationinfrastructures that are more flexible, efficient, androbust in order successfully operate in the face of enemyattacks.Most communications between peers in the battlefield areeither to exchange state and environmental information orto relay command and control messages. State informationincludes, for instance, relative position of troops and vehicles(enemy and friendly), sensor data from unmanned vehiclesor sensor beds, situation data, etc. This type of datais often transmitted as streams of arbitrary durations, suchas video-feeds from a camera sensor or continuous GPSposition data from moving vehicles or troops.Furthermore, the communications infrastructure must adjustto changes in overall mission goals and operationstempo. During monitoring and recognition missions, conservingpower might be the primary objective function toextend the life of network resources. However, as engagementtakes place, the communications infrastructure mayneed to quickly shift into a high performance mode to effectivelysupport and optimize the kill chain as the primaryobjective.In this paper we present FlexFeed, a mobile-agent basedcommunications framework, applied to the battlefield scenario.Our proposal leverages from years of research in thefields of mobile ad hoc networks and intelligent softwareagents to build an efficient, self-configurable, and selfhealingcommunications network for these types of environments.After an introductory description of the environment andsystem requirements, we will discuss the related work inthis area and the concepts proposed in FlexFeed. A briefdescription of the implementation details of frameworkwill be then followed by case studies, presented and experimentallyevaluated on a simulated network to illustrateFlexFeed capabilities.COMMUNICATIONS IN THE BATTLEFIELDAs part of the Army’s Objective Force to be deployedwithin the next decade, Future Combat Systems (FCS) isenvisioned as a system of systems that will integrate severallightweight, highly mobile components including newgenerations of manned and unmanned military vehicles, hese light vehicleswill partially replace heavy armored slower vehiclesin order to bring unprecedented levels of dynamism andagility to the combat theater.Furthermore, FCS operations will heavily rely on informationsuperiority to quickly take control of the battlefield and react to enemy movements and changesof strategy. This capability depends on the notion of universaltasking, where resources and information are directlyavailable at any timeto the edge warriors and commandersin the field.An enabling key-capability for this vision is an efficientand adaptive communications infrastructure to support andextend edge warrior capabilities and provide access tocritical information at any time, while at the same timeensuring optimal resource utilization and security both atthe infrastructure and information levels. Figure 1 shows aschematic view of some of the elements involved in theseeypes of operations.In general, a communications model capable of supportingFCS requirements is an ad-hoc publish/subscribe model.Soldiers and systems in the network will subscribe to sen2of 8 data and state information to plan and coordinate localtasks in response to high level instructions from the commandand control center.Figure 1 –Communications infrastructure in the battlefieldBased on FCS requirements, an appropriate data communicationsframework must be capable of satisfying the followingrequirements:a) Ad hoc: In most cases, as illustrated in figure 1, networksbetween nodes will be ad hoc, formed by proximityduring the operation itself. The communications infrastructuremust not depend on pre-established infrastructuralcomponents or centralized management stations. This capabilityis important both from a scalability and robustnessperspective, eliminating (or mitigating) single points offailure in the network.b) Efficiency: Communications and computational resourcesin the battlefield are expected to be limited andoften times, battery operated. Ad hoc sensor beds andsmall autonomous vehicles deployed during the operationwill have a life-span strictly limited by their battery life. Inmost cases, it is imperative that the communications infrastructure operate efficiently across different types of applicationsand scenarios to extend the life of network resources.c) Heterogeneity: Systems in the battlefield tend varygreatly in terms of computation and communications capabilities.Lightweight attack vehicles, small robotic units,and unmanned aerial vehicles will all have different degreesof computation and sensing capabilities andaccess tothe wireless environment.d) Application-aware Capabilities: A common limitation inmost communications frameworks currently available isthe lack of interaction between applications and theunderlyingdata transmission protocols. This limitation is oftenaccepted in lieu of the benefit of layer isolation and inmaking protocols interchangeable. For improved efficiencyhowever, the communications infrastructure canand should benefit from data-aware protocols at all levels.d) Robustness to External Attacks: The communicationsinfrastructure must be able to resist to both physical andnetwork attacks. Degradation with loss of communicationresources must be graceful and most importantly, must beselective. Special types of operations and tasks that arecritical to the overall operation must have precedence overless relevant tasks. This requirement goes beyond theconventionalnotion of quality of service in data networks.Ideally, the framework must be aware of the importance ofdata transmission not only in terms of data-type, source,and destination, but also in terms of high level goals andmission OPTEMPO in order to makeprioritization decisions.c) Robustness to Environmental Changes: The environmentalconditions, topology, and size of the network willvary significantly. In the battlefield, nodes can arbitrarilyjoin and leave the network. Nodes can be physically destroyedor made unavailable at any time. Anappropriatecommunications infrastructure must be able to cope withthese changes quickly and efficiently.e) Reactive and Proactive OPTEMPO Adaptability: Theframework must also be able to properly adapt to changein overall mission goals or situation in the battlefield. Forinstance, changes in operational tempo can be eitherpushed to or autonomously detected by framework nodes,which should automatically result in changes in the communicationsbehavior. Forinstance, when precursors ofengagement are identified, the framework, in accordancewith global policies, must autonomously switch from apower efficient mode a low latency, high performancemode to support combat systems.f) Proactive Resource Manipulation for Survivability andImproved Efficiency: This notion was initially proposedwithin the context of Agile Computing (Suri, 2002). It refersto the notion of granting the framework with the abilityto proactively manipulate physical (or logical) resourcesin the framework in order to recover criticalconnectivity segments or to significantly improve performance.g) High Level Policies for Monitoring and Control: From ahuman perspective, monitoring and control of such complexsystems is a very difficult task. An appropriate frameworkfor these types of systems must support interfaces topolicy infrastructures that would allow humans to easilydefine and establish constraints and obligations to regulatethe overalloperation of the framework. From an optimizationperspective, most policies would ultimately result inlow level constraints taken into account by the frameworkwhen deciding about resource allocation.In the last few years, a number of research proposals havebeen introduced to address some of these mon to most of them is the notion of a customizablepublish/subscribe communications mechanism capable toefficiently support messaging and data streaming.RELATED WORKConventional topic-based publish/subscribe systems suchas such as Vitria (Skeen, 1998), TPS (Eugster et al., 2001)3 of 8and JORAM (Maistre, 2003) leveraged form multicast protocolsand the assumption of a clear hierarchy on data andevents to build efficient multicast groups for topic-baseddata distribution. Multicast based protocols often providean efficient solution to the problem but they assume thatonly nodes participating in the multicast group wouldshare the data for distribution (at the level of the multicasttree). Furthermore, most multicast protocols assume data(or events) to be strictly hierarchical and processing capabilitiesfor data transformation within the hierarchy mustbe available a priori at all nodes.A number of gossip-based (or epidemic) protocols werealso proposed in the same context (Lin and Marzullo,1999; Ganesh et al., 2001; Eugster et al., 2003). In general, these are efficient and scalable protocols but assume nodata hierarchy and often make no attempt for cost or constraintoptimization based on data stream aggregationandfiltering.Multicast protocols specifically designed for peer-to-peernetworks such Scribe (Rowstron et al., 2001) and HiCan(Ratnasamy et al., 2001) came to solve scalability issues inaddressing and group coordination. They too, however,assumed that only nodes subscribedto the multicast groupwould participate in the multicast tree and that data processingcapabilities are available at all nodes a priori.Alternatives to the multicast option were also proposed atthe level of unicast routing in the form of data-aware customizedad hoc routing protocols. An important exampleof these types of data-centric routing protocols is DirectedDiffusion (Intanagonwiwat et al. 2000). The Directed Diffusionprotocol proposes a highly scalable data-aware decentralizedrouting algorithm. The protocol supports thecreation of data distribution trees including nodes that arenot directly subscribing for the data. The protocol, however,also assumes that data transformation capabilities areavailable a priori at each node, which is not a realistic assumptionfor the types of environments envisioned in FCS.More recently, Baehni et al. (2004) proposed a data-awaremulticast protocol (daMulticast) for peer-to-peer networks.The approach leveraged from some of thedata-centrictechniques for data description and group membership,significantly improving reliability and at the same timereducing the memory complexity involved in maintaininggroup membership at each node.In the most part, the approaches share the notion of usingdata-aware techniques for resource or performance optimization.The problem, however, is that data-aware frameworksare usually highly customized to a set of applicationsor data types, often requiring significant time andeffort to support new scenarios or capabilities. In manycases, such changes are not even possible, as hardwaremight have been already deployed or might be under externaladministrative control, like in the case of combat orMilitary Operations Other than War (MOOTW) coalitionoperations.THE FLEXFEED FRAMEWORKIn this paper we propose FlexFeed, a mobile-agent basedcommunications framework designed to support highlycustomized data streams in mobile ad-hoc network environmentsunder policy and resource constraints.The concepts implemented in FlexFeed were first introducedby Carvalho et al. (2002). The fundamental ideas ofthe framework are based on the concepts of Agile Computing(Suri, 2002) where network and system resources areopportunistically exploited to transparently support applicationrequests in a manner that is efficient, robust, andadaptable to changes in the environment.The FlexFeed framework is essentially based on three coreconcepts: a) Opportunistic resource exploitation; b) Flexibilityand run time self-configuration via on-demand codeand process migration; and c) In-stream data processing.Inthe framework, these capabilities are combined and extendedto address the requirements identified in the typesof environments expected in FCS.The framework uses data-aware mobile agents to bettercustomize multicast trees and to provide in-stream dataprocessing (i.e. to take advantage of the multi-hop natureof the communications path in these types of environmentsto distribute computation data processing loads). Specializedagents can be injected in the framework by authorizedparties at run-time, allowing for great flexibility and supportof highly specialized data streams. The overall behaviorof the framework is regulated by high level policiesdefined, verified, and distributed by an integrated policyinfrastructure designed for multi-agent systems. A proof-concept version of the FlexFeed framework was developedand tested both in simulated and physical environments.The framework was also demonstrated in actuallive exercises conducted by theArmy (ARL QL2, 2004)and the demonstrations for the Navy (ONR NAIMT,2004). In the subsequent items, we will briefly discuss theimplementation details of the framework, followed by experimentalsimulation results of illustrative case studies.THE FRAMEWORK COMPONENTSThe FlexFeed framework is a distributed application-leveliddleware that is installed in all participating systems.te middleware provides an API that allows applications specify services or requests for data streaming.At heimplementation , the framework combines amobile agent system with resource coordination and allo4of 8cation mechanism and a policy infrastructure to determined configure, at runtime, efficient data distribution treesbetween applications.The mobile agent system gives the framework the abilityto move code and computation between nodes to enable,on demand, new data-specific capabilities in nodes thatwill participate in the data distribution tree. Process migrationis used to improve survivability and system performance.AlthoughFlexFeed can be easily configured to workwith different agent systems, our proof of concept implementationwas developed on top of the NOMADS agentsystem (Suri et al., 2000; Groth and Suri, 2000).NOMADS is a mobile agent system for Java-based agents.It provides two implementations: Oasis and Spring. Oasisincorporates a custom Java-compatible Virtual Machine(named Aroma) whereas Spring is a pure Java implementation.The Aroma VM is a clean-room VM designed toprovide the enhanced capabilities of execution state captureand resource control.The resource coordination component (referred to in thispaper as the ‘coordinator’) is the intelligent part of theframework. It is responsible for realizing the notion of agilecomputing in the context of data streaming. The ‘coordinator’can be implemented as a distributed process or asa centralized component operating in one of the nodes ofthe framework. All experiments and examples shown inthis paper are based on one specific implementation of acentralized coordination algorithm (ULM) but decentralizedalternatives are also available.The policy infrastructure is independent of the framework.The goal of the policy framework is to provide a high levelinterface to the system in order to allow both human operatorsand applications to establish, query and modify highlevel requirements and constraints that will regulate howthe framework should operate. Furthermore, the policyinfrastructure is also responsible for validation, verification,disambiguation, and distribution of policies throughoutthe system. Policies can also be used to regulate andconstrain the autonomous behavior of the framework, providingbounds for self-adjustments to operation tempo andto the proactive manipulation of resources. Currently,FlexFeed uses KAoS (Bradshaw et al., 1997; Bradshaw etal. 2002; Bradshaw et al., 2003) as its policy framework.Access to these components is available at each nodethrough a common API. In order to participate in theframework, applications at each node can obtain an instanceof the FlexFeedManager Component (Figure 2).The FlexFeedManager provides the access API to the framework and allows applications to register, advertisecapabilities, and request data streams from other resources.Transparent to the applications, FlexFeedManagers at communicate in a peer-to-peer fashion to exchangestate and plan resource utilization. When a client places arequest for a data stream from a sensor as illustrated infigure 2, it specifies the source of the data and the requirementsfor the data stream (for example, resolutionand frame rate in the case of a video stream).That information, along with resource availability informationfrom local nodes, is used to build the data distributiontree from source to client. If using a centralized coordinator,the planning is done at one single location using globalstate data. Decentralized coordination algorithms rely onpeer to peer negotiation between FlexFeedManagers anduse only local state for planning.The client is allowed to specify any type of data, grantedthat it provides to the framework the necessary informationfor cost calculation and the code (in the form of mobileagents) necessary to manipulate (e.g. aggregation and filtering)the data for optimization. Because FlexFeed supportson-demand code deployment, trusted applicationscan provide new components to the framework at run time,enabling the support of previously unknown data types.The client can also provide complex data processing requestssuch as the one illustrated in figure 3. In that example,the client is specifying (through a graph structure) twodistinct data sources that should be merged with a specific(client-provided) processing element (FS) and then, discriminativelydelivered to two sink nodes. Details aboutthe data types and processing elements are embedded inthe graph node and edge components, using a pre-defineddata structure provided by the framework.The FlexFeed framework will load the appropriate softwarecomponents specified by the client (either from theclient host of from a common codebase) and will identifythe network resources necessary to support the request.The location of the logical fusion element (FS) can be atany intermediate node between the source and sink elements,based on resource availability, policies, and overallcosts for computation and data transmission.After mapping the request to the physical network, theframework will monitor environmental changes (such assignificant variations in resource availability or link failure)to transparently recalculate and adjust the data treeuntil the request is terminated by the client.CASE STUDIES AND EXPERIMENTAL RESULTSThe framework was tested in a simulated network wherepacket drops and bandwidthconstraints could be carefullycontrolled. The goal of these tests was to demonstrate theeffectiveness of on demand configuration of data-awarestreaming in the improvement of data quality and reductionof jitter. These metrics are highly relevant to applicationssuch as the remote control of unmanned vehicles.The overhead of the framework was also measure in termsof induced latency in the stream. The computational overheadfor running intermediate processing elements andfilters was disregarded and the coordination mechanismused in the experiments was the centralized ULM (Carvalho,2005) algorithm, based on an iterative version ofDijkstra’s shortest path algorithm applied in localized partsof the graph.The experiments were conducted on a 100baseT networkwith full connectivity. Bandwidth limitations betweenUA V and other nodes were simulated on a fixed wirednetwork. Figure 4 provides a schematic view of the testnetworks.Figure 4 – Schematic illustration of the environment consideredfor experiments In this configuration, nodes ‘S1’ and ‘S2’ represent two dismounted soldiers in direct communications range witheach other and with a tank nearby ‘T2’. All nodes are incommunications range with an unmanned aerial vehicle‘UA V’ on a fixed flying pattern over enemy territory.The bandwidth available from the UA V to the remainingnodes is variable and can be severely constrained at differenttimes. Our experimental procedure explores differentoperational scenarios on top of this configuration. The goalis to quantitatively illustrate how the FlexFeed frameworkimproves data communications by reducing delays betweenimage updates and the variance between packet arrivaltimes (jitter). In our experimental setup, each node isrepresented by a separate laptop. The bandwidth limitationson the UAV are simulated by a gateway runningNISTNet (Carson, 2002). Figure 5 shows the experimentalsetup designed to simulate the environment illustrated infigure 1.The centralized coordination node (not shown in figure 5)receives state information such as CPU and bandwidthavailability from each of the 4 nodes involved in the test.The frequency of updates is proportional to the rate ofchange in these metrics. When a client makes a request fora data stream, it specifies the source node (UA V), framerate, and resolution. The coordinator nodewill receive therequest and will handle it appropriately, building a datadistribution tree from the source, based on current globalsystem state.Optimizing Bandwidth UtilizationIn the first scenario, soldier ‘S1’ temporarily assumes controlof the UA V, taking it o ut of the flying pattern andcloser to enemy positions. Unaware of the fact that the vehicleis now under remote control, soldier ‘S2’ also requestsa video stream from the UA V’s camera. In this example,both video streams were requested at a 320x240resolution with 3 frames per second.Under unconstrained conditions, the combined streamsrequire the UA V to send approximately 50 KBps of data.In the initial condition the bandwidth limitation is 100KBps (equivalent to unconstrained bandwidth in this example)so there are no packet drops and the average delaybetween images is 271 milliseconds, which represents astream of approximately 2.69 frames per second. Note thatthe resulting frame-rate, even under unconstrained bandwidthconditions, only approximates the requested framerate.This is due to the delays involved in actual imagecapture (which is camera dependent), compression, andserialization.The bandwidth available from the UA V is then progressivelyreduced to a maximum of 40 KBps, 30 KBps, and6 of 8then 24 KBps. At each step, the average delay betweenimages is measure at each client ‘S1’ and ‘S2’.When the coordinator is inactive, that is, when the frameworkis not making any attempt to optimize data streams,the sensor (UA V) sends a unicast stream to each of theclients. Both streams will compete for the limited bandwidthand the delays at each client increase significantlywith the reduction in bandwidth availability. These resultsare shown in figure 6, with their 95% confidence errormargins.Figure 6 – Effects of bandwidth reduction without datastream coordination.In this example, as the bandwidth availability decreases,the delays quickly increase to the point where criticaltasks, such as the remote control of the UA V, are completelycompromised. A minimum frame-rate of 2 fps isrequired1, in this example, to safely navigate the UA V so itis clear that even small constraints in bandwidth availabilitycan compromise this task. Furthermore, we can verifythe well known bandwidth stealing behavior between clients,where bandwidth is not equally shared betweenstreams. This behavior has been previously reported in IPnetworks (Tschudin and Ossipov, 2004) and could compromisecritical tasks such as the control of the UA V.Figure 4 – Data distribution tree created by FlexFeedWhen the FlexFeed coordinator is enabled, the frameworkidentifies the stream requests and attempts to globally optimizedata distribution. In this specific case, the coordinator(which, is a centralized process) determines that bothstreams are similar (in fact, equal) and the overall bandwidthutilization can be reduced with a multicast-like datadistribution tree. The framework opportunistically identi-1 Although it is commonly accepted that a minimum of four frames persecond is necessary to remotely operate robotic vehicles, for illustrationpurposes in this example, the minimum requirement for teleoperationis assumed to be two frames per second.fies node ‘T2’ as a potential intermediate processing elementand builds the distribution tree illustrated in figure 4.Under the same bandwidth constraints, the framework ensuresthat the lowest capacity link (from the UA V) is notsaturated and delays between images are kept within reasonablebounds.Furthermore, the variance in delay (jitter) is significantlysmaller, ensuring that critical processes maintain minimum levels of throughput and quality of service.FlexFeed OverheadThe overhead of framework basically falls into two maincategories: a) the number of additional control messagesinvolved on sharing state between nodes (or betweennodes and the centralized coordinator) and b) the time requiredto determine, locate, and configure the nodes thatwill participate in the data stream. Both factors are highlydependent on the type of coordination mechanism used inthe framework (centralized, zone-based, or local), the complexityof the data, the scale of the network, its level ofconnectivity, and the frequency of state updates.In our example, the network topology is static and variationsin resource availability are small so our attention isfocused primarily on the delays (latency) caused by thecentralized coordination algorithm. Table 1 shows the averagedelays and their 95% confidence error margins observedin each test, both with and without the coordinator.The delays were measured as the average between the timeof the second client ‘S2’ request and the time when thefirst image is delivered to that client.When the coordinator is present (second column), there isan up-front cost in terms of latency that is due to the timespend in identifying and configuring network resources fordata distribution. When the coordinator is not present, theresponse to the data request is relatively fast at first but thedelays increase as the bandwidth is reduced. This is basicallydue to the fact that initial images are being lost on thesaturated channel when the coordinator is not present.FlexFeed versus MulticastThe data distribution tree presented in this example resemblesa data multicast tree, often obtained with conventionaldata multicast algorithms. As previously noted, FlexFeed7 of 8goes beyond conventional and data-aware multicast approachesby building a tree that include nodesthat are notnecessarily part of the multicast group (node T2 in thiscase). These nodes are opportunistically discovered andconfigured, at runtime by the framework, based on its currentresource availability, role in the network, and systempolicies.Furthermore, the configuration of node T2 can be highlydata-dependent and arbitrarily complex. In these examples,the intermediate node was use merely as a splitting pointfor the data distribution tree. It could also have receivedcustomized code to perform in-stream data transformationor specialized filtering.Consider the case where the request place by soldier ‘S2’was for a lower res olution video stream from the samesource. Multicast algorithms would often regard this as anindependent request or would have assumed that one of thenodes in the multicast group would be able to reduce theresolution of the stream to include the new request in thedata hierarchy. Data-centric protocol like Directed Diffusionwould also depend on an intermediate node’s a prioricapabilities to construct the low resolution data from thehigh resolution stream.In FlexFeed, the request placed by soldier ‘S2’ can sp ecifyreferences to data-specific code that will be installed, ondemand, on node ‘T2’ to act as a processing element. Thecode would be installed only in the necessary nodes (asdetermined by the coordination algorithms) and would beremoved when no longer necessary.Another extension of the same capability is the transparentenforcement of information release policies. In this case,upon S2’s data request, FlexFeed would query the policyframework for constraints or obligations involving the request.Consider, for example, that policies were previouslydefined to constrain unrestricted access from S2 to thatspecific data source. In that case the framework will,transparent to node ‘S2’, identify an intermediate node forpolicy enforcement and will deploy the customized datafilters (specified as part of the policy) to ensure compliancewith the specified requirements. This feature of Flex-Fleed has been extensively demonstrated by (Suri, Bradshawet al, 2003; Suri, Carvalho et al, 2003) in multiplesimulations and real life exercises.CONCLUSIONS AND FUTURE WORKIn this paper we have introduced the FlexFeed frameworkin the context of military combat operations. The conceptproposed in FlexFeed goes beyond current data-centricrouting approaches and data-aware multicast. It realizesthe notion of agile computing in the context of data communicationsand offers the basis for a truly customizablemiddleware for data communications in extreme environments.The framework is currently implemented and has beentested in several small scale exercises including soldiersoperating in conjunction with robotic units and remote systems.The framework currently relies on a centralized coordinationalgorithm for resource allocation. We are currentlydeveloping fully decentralized (and zone-based)algorithms to improve scalability, robustness, and performance.。
软件工程博士点主要研究领域、特色与优势英文版Main Research Fields, Features, and Advantages of Doctoral Programs in Software EngineeringSoftware engineering, as a discipline, has witnessed significant growth and transformation over the years, evolving from a mere focus on coding to encompassing various areas like system design, project management, and software quality assurance. Doctoral programs in software engineering aim to cultivate experts who not only possess profound technical knowledge but also have the ability to innovate and lead in this rapidly evolving field.1. Main Research FieldsDoctoral programs in software engineering typically cover a wide range of research areas. Some of the key fields include software development methodologies, software architecture, artificial intelligence and software engineering, software testingand quality assurance, and distributed and cloud computing. These fields aim to address the complexities and challenges associated with software development in today's interconnected world.2. FeaturesThe features of these doctoral programs are often unique and tailored to meet the needs of the industry. They强调rigorous theoretical training combined with practical experience, often through internships or research projects. The programs also focus on fostering critical thinking, innovation, and leadership skills, enabling graduates to become agents of change in the software engineering landscape.3. AdvantagesThe primary advantage of pursuing a doctoral program in software engineering is the depth and breadth of knowledge gained. Graduates are equipped with a solid understanding of the theoretical foundations of software engineering, enabling them to contribute significantly to research and development inthe field. Additionally, they gain exposure to cutting-edge technologies and trends, making them highly competitive in the job market.The doctoral programs in software engineering prepare individuals for roles that demand not just technical proficiency but also strategic thinking and leadership. The comprehensive nature of these programs, coupled with their focus on innovation and research, ensures that graduates are well-prepared to take on challenges and opportunities in the rapidly evolving world of software engineering.中文版软件工程博士点主要研究领域、特色与优势软件工程作为一门学科,多年来经历了显著的发展和变革,从仅关注编码扩展到了系统设计、项目管理、软件质量保证等多个领域。
Software Agentsfor Market Design and AnalysisFinal Project ReportPower Systems Engineering Research CenterA National Science FoundationIndustry/University Cooperative Research Centersince 1996Power Systems Engineering Research Center Software Agents for Market Designand AnalysisProject TeamSarosh Talukdar, Project LeaderCarnegie Mellon UniversityTim Mount, Robert J. ThomasCornell UniversityShmuel OrenUniversity of California at BerkeleyPSERC Publication 05-37June 2005Information about this projectFor information about this project contact:Sarosh N. TalukdarProfessorElectrical and Computer EngineeringCarnegie Mellon UniversityPittsburgh, PA 15213-3891Phone: 412-268-8778Email: talukdar@Power Systems Engineering Research CenterThis is a project report from the Power Systems Engineering Research Center (PSERC). PSERC is a multi-university Center conducting research on challenges facing a restructuring electric power industry and educating the next generation of power engineers. More information about PSERC can be found at the Center’s website:.For additional information, contact:Power Systems Engineering Research CenterCornell University428 Phillips HallIthaca, New York 14853Phone: 607-255-5601Fax: 607-255-8871Notice Concerning Copyright MaterialPSERC members are given permission to copy without fee all or part of this publication for internal use if appropriate attribution is given to this document as the source material. This report is available for downloading from the PSERC website.©2005 Carnegie Mellon University. All rights reserved.AcknowledgementsThe work described in this report was sponsored by the Power Systems Engineering Research Center (PS ERC). It is the final report for the PS ERC project “Bidding Agents”. We express our appreciation for the support provided by PS ERC’s industrial members and by the National Science Foundation under grant NSF EEC-0109543 at Carnegie Mellon University, and grant NSF EEC-0118300 at Cornell University received through the Industry/University Cooperative Research Center program.Executive SummarySoftware agents can be of help in the design, analysis and verification of markets. The agents described in the four chapters of this report do two things. First, they identify some of the uses of software agents. Second, they show that competitive behavior is not guaranteed for existing electricity auctions. Rather, suppliers can drive prices to well above competitive levels and make excessive profits.Chapter-1 covers software agents designed to mimic test-results obtained from human subjects in simulated electricity markets. The agent behavior is compared to the behavior of the subjects. Both human and software-agent behavior are classified based on the data. Differences and similarities are noted and explained. The agents are used to demonstrate that the hope that large-scale generating units will operate at marginal cost in a uniform price auction, is at best wishful thinking. In fact, both real and experimental data show that the more uncertainty a supplier faces (e.g., load uncertainty, uncertainty of other suppliers, etc.) the more the supplier will try to increase its profits by submitting offers to sell at higher than marginal cost and by withholding units, if permitted. This makes predicting unit commitment and dispatch ahead of time difficult.Chapter-2 covers the use of a multi-agent system (MAS) to simulate a spot market with several supply firms. The firms submit offers to maximize their own expected profits, and an Independent System Operator (ISO) clears the market for a predetermined load in a uniform price auction. The firms learn about the structure of the market and the behavior of their competitors by comparing actual market outcomes with predicted outcomes based on an estimate of their own residual demand curve. This estimated demand curve is updated each period using a Kalman filter. The simulations were used: 1) to determine which characteristics of a deregulated wholesale market for electricity make price spikes likely to occur, and 2) to determine a structure of firms, in terms of size and type, that replicates the infrequent price spikes observed in the PJM market during the summer of 1999. The two most important characteristics of the market for creating price spikes are 1) uncertainty about the system load is the primary determinant of the observed speculative behavior (i.e. submitting offer curves shaped like a hockey stick), and 2) all firms eventually become speculators, and it is unrealistic to expect firms to behave like price takers in a market with only a few suppliers. In other words, given the intrinsic uncertainty about the load (and the possible outages of generators), speculative behavior by suppliers is entirely rational. Overall, the results show that the analytical framework of a MAS can replicate observed price behavior in a deregulated electricity market, can provide new insights into how suppliers behave, and has the potential for evaluating a wide range of policy options for mitigating high prices.Chapter-3 investigates how spot prices are affected by forward contracts. Experimental economics and agent-based simulations are used. The results show that holding a forward contract is an effective way to mitigate high prices if the same contract is held for all trading periods and the price of this contract is independent of the spot prices. However, when a forward contract is renewed periodically and the spot prices influence the forwardprice, there is more speculation and the spot prices may be higher than the base case with no forward contracts. The simulation results also show that software agents were able to replicate the behavior of students effectively in the experiments.Chapter-4 uses software agents to examine repeated auctions. The purpose of the agents is not to mimic the behavior of humans. Rather, it is to reveal those fundamental properties of repeated auctions that are independent of human idiosyncrasies; just as wind tunnels and finite element programs are used to reveal the fundamental properties of new airplanes before test pilots are made to fly them. Simple experiments show that existing designs of repeated auctions have at least one major flaw: The sellers, even though they work without collusion, can learn strategies that raise prices and profits far above competitive values. This happens even when the demand is price-responsive. A remedy is to make the auctions symmetric, that is, to provide the buyers with as much autonomy as the sellers have, so the buyers, or their surrogates, can learn, and otherwise make decisions as quickly and cleverly as the sellers. However, this remedy is unlikely to be the only one needed. Complex artifacts, unless their designs are thoroughly verified, invariably suffer from major flaws. Thorough verification requires a set of tests that spans both the operating conditions and the desired behaviors of the artifact. What passes for verification in market design—opinion surveys and a few random experiments with human subjects—is far from adequate.Acknowledgments (i)Executive Summary (ii)Table of Contents (iii)Table of Tables (v)Table of Figures (vii)1. Comparing the Behavior of Agents to Human Subjects in a Uniform Price Auction.............................................................................................................................1-1 1.1 Introduction ..............................................................................................................1-1 1.2 Electricity Market ....................................................................................................1-2 1.3 Standardized Agents ................................................................................................1-2 1.4 Classification of an Agent ........................................................................................1-4 1.5 Expected Earnings ...................................................................................................1-5 1.6 Simulation Results ...................................................................................................1-6 1.7 Conclusions ............................................................................................................1-13 1.8 Future Work............................................................................................................1-13 1.9 Acknowledgment ...................................................................................................1-131.10 References .............................................................................................................1-142. Using Software Agents to Replicate Observed Price Behavior in the PJM Electricity Market............................................................................................................2-1 2.l Introduction ...............................................................................................................2-2 2.2 A Multi-Agent System for Testing Electricity Markets ..........................................2-3 2.2.1 The Model of a Deregulated Electricity Market ...................................................2-3 2.2.2 Using Adaptive Autonomous Agents for Firms ...................................................2-4 2.2.3 Limitations of Deterministic Profit Maximization ...............................................2-5 2.3 Creating Price Spikes in a MAS ..............................................................................2-7 2.3.1 Salient Features of the Market ..............................................................................2-7 2.3.2 Speculative Behavior is Almost Inevitable ...........................................................2-8 2.3.3 Speculation Increases when Load is Uncertain ..................................................2-11 2.4 Replicating Observed Price Behavior in the PJM Market .....................................2-12 2.4.1 Using the Observed Pattern of Load in PJM ......................................................2-12 2.4.2 Too Many Firms are Required to Make Textbook Competition Practical ........................................................................................................................................2-13 2.4.3 The Effect of Including Vertically Integrated Firms is Relatively Small ................... ........................................................................................................................................2-14 2.4.4 Replicating the Structure of firms in PJM ..........................................................2-15 2.5 Conclusions ............................................................................................................2-17 2.6 References ..............................................................................................................2-19(continued)3. Using Software Agents to Supplement Tests Conducted by Human Subjects: How Do Forward Contracts Affect Behavior of Suppliers in a Wholesale Electricity Market? ...........................................................................................................................3-1 3.1 Introduction ..............................................................................................................3-1 3.2 Optimum Offer Behavior in a Spot Market .............................................................3-3 3.3 A Supplier’s Expected Profit ...................................................................................3-3 3.4 The Optimum Offer .................................................................................................3-4 3.5 The Experimental Framework .................................................................................3-6 3.6 Analysis of the Test Results .....................................................................................3-8 3.7 The Performance of Computer Agents ..................................................................3-12 3.8 Summary and Conclusions ....................................................................................3-153.9 References ..............................................................................................................3-174. Symmetry and Verification : Critical Parts of Market Design ..................................4-1 4.1 Introduction ..............................................................................................................4-1 4.2 The Dynamics of Repeated Electricity Auctions .....................................................4-2 4.3 Overview of the Experiments ..................................................................................4-4 4.4 Experiment-1: Price-Invariant Demand ...................................................................4-6 4.5 Experiment-2: Price-Responsive Demand ...............................................................4-9 4.6 Experiment-3: Cooperation ....................................................................................4-10 4.7 Experiment-4: Active Demand ..............................................................................4-11 4.8 Conclusions ............................................................................................................4-15 4.9 Acknowledgement .................................................................................................4-17 4.10 References .............................................................................................................4-17Table 1.1 Offer strategies of the standardized agent when the fairshare block is thejth block. Here MCO, S and W stand for the marginal cost offer, speculate and withhold, respectively. ....................................................................................................1-4 Table 2.1 Optimum Solutions for Profit Maximization by a Firm ................................2-5 Table 2.2 The Number of Firms at the Beginning and End of a Simulation by their Belief about High Prices ...............................................................................................2-10 Table 2.3 Average Price for Different Numbers of Firms (Competitive price is $39/MWh) .....................................................................................................................2-13 Table 2.4 Average Price for Different Numbers of VIF (Competitive price is $39/MWh) ......................................................................................................................2.15 Table 3.1 The Optimum Offers in the Spot Market .......................................................3-5 Table 3.2 Impact of Contract Quantity on Optimum Offers ..........................................3-6 Table 3.3 Regression Results .......................................................................................3-10 Table 3.4 Average Earnings ($ in periods 11 to 20) ....................................................3-11 Table 3.5 Average Market Prices ($/MWh) ................................................................3-12 Table 3.6 Average Earnings ($MWh)* for All-Agent Tests .......................................3-15 Table 3.7 Average Market Prices ($MWh)* for All-Agent Tests ...............................3-15Figure 1.1 Schematic diagram showing the offer functions of the different types of standardized agents..........................................................................................................1-4Figure 1.2 Earnings of the standardized agents and the agent of interest.......................1-5Figure 1.3 Example of a performance of the software agents: in the plot, red square, green and blue circle stand for the earning in a period of MC, WS, and SS, respectively......................................................................................................................1-7Figure 1.4 Example of the performance of human agents: the red square, green and blue circles represent a single period earnings of MC, WS and SS agents, respectively. ..................................................................................................................1-10 Figure 1.5 Actual earnings vs. expected earnings calculated from a simulation. ........................................................................................................................................1-12Figure 2.1 A Multi-Agent System for an Electricity Spot Market..................................2.4Figure 2.2 Simulated Market Price.................................................................................2-8Figure 2.3 Load and Price for Case E3/S5......................................................................2-9Figure 2.4 The Effect of Different Load Forecasting Errors (0% and 5%) on the Optimum Offers (OPT) of a Firm (MC is the Marginal Cost)......................................2-11 Figure 2.5 The Probability of Rejecting an Offer in the Auction (5% Forecasting Error)..............................................................................................................................2-12Figure 2.6 Actual Load and Peak Price in the PJM Market (April to September, 1999)..............................................................................................................................2-13 Figure 2.7 Actual Market Prices and Simulated Market Prices Using Naïve Agentsfor Firms.........................................................................................................................2-16Figure 2.8 Simulated Market Prices Using Naïve and Trained Agents for Firms ........................................................................................................................................2-16 Figure 3.1 Optimum Offer Price with Contracts............................................................3-5 Figure 3.3 Comparison of Average Earnings...............................................................3-13(continued)Figure 4.1 The rows in the matrix, A kM, represent offers from the sellers for the k-th auction; a “1” denotes a generator whose energy is offered for sale, a “0” denotes a generator that is withheld. The pattern of 1’s and 0’s in A kM produces the maximum total profit for the sellers, $3860, just as they would obtain if their generators were controlled by a monopolist with perfect information, including perfect predictions of demand.............................................................................................................................4-3Figure 4-2 Twelve demand curves of which curves 1 - 3 are price-invariant, 4-12 are price-responsive...............................................................................................................4-7Figure 4.3 Aggregate supply curves from Experiment-1 with the demand fixed at 50 MW. Initially, all the suppliers offer all their generators at cost. The resulting aggregate supply curve is represented by the broken line and the clearing price in the first auction is the competitive price: $3.13/MWh. As the auctions proceed, the suppliers learn to withhold generators. At repetition 1000, the withholding is so effective that the aggregate supply curve (solid line) intersects the demand at the maximum possible price, $100/MWh. ............................................................................4-8Figure 4.4 A trajectory of clearing prices from Experiment-1, with the demand fixed at 50 MW. The fluctuations in the trajectory decrease as the trajectory develops, but never completely disappear because the learning algorithm continues to makes random changes in the suppliers’ offers. By repetition 1000, the average clearing price (averaged over 50 consecutive repetitions) has become $98/MWh, and it remains at this value thereafter. ...............................................................................................................4-9Figure 4.5 Profits and learning yields from Experiments-1 and –2. Whether the demand is price-invariant (demand curves 1-3) or price-responsive (demand curves 4–12), the sellers are able to obtain between 72% and 87% of the monopolistic profit. ...............4-10Figure 4.6 Profits and learning yields from Experiment-3, in which sellers cooperate by a) exchanging data on their individual profits, and b) seeking to maximize total, rather than individual, profit. ...................................................................................................4-11 Figure 4.7 Some results from Experiment-4. Both the suppliers and the buyers are allowed to learn. Consequently, both the offers and the bids change from one repetition to the next. The broken curves represent the competitive supply and the actual demand. The solid curves show the supply that is offered and the demand that is bid at repetition-1000. .............................................................................................................4-12(continued)Figure 4.8 More results from Experiment-4. The broken lines represent the supply and demand curves at repetition-1. (The demand corresponds to Curve-10 in Fig 2). The solid lines show these curves in repetition-1000. The changes are learned. .................4-13Figure 4.9 A trajectory of clearing quantities from Experiment-4. with the demand corresponding to of Curve-4 in Fig. 2. ..........................................................................4-14Figure 4.10 A clearing price trajectory from Experiment-4 with the demand corresponding to Curve-4of Fig.2. ...............................................................................4-151. Comparing the Behavior of Agents to Human Subjects in a UniformPrice AuctionAbstractThe idea that large-scale generating units will operate at marginal cost when given the ability to offer their power for sale in a uniform price auction is at best wishful thinking. In fact, both real and experimental data show that the more uncertainty a supplier faces (e.g., load uncertainty, uncertainty of other suppliers, etc.) the more they will try to increase their profits by submitting offers to sell higher than marginal cost and by withholding units if permitted. This makes predicting unit commitment and dispatch ahead of time difficult. In this Chapter we explore characteristics of software agents that were designed based on the outcome of tests with human subjects using a uniform price auction with stochastic load. The agent behavior is compared to the behavior of the subjects. Both subject and agent behavior are classified based on the data. Differences and similarities are noted and explained.1.1. IntroductionFor the past decade, deregulation in the electric power market has been taking place in many countries. In many deregulated markets, an auction plays a major role in determining the price for electricity, using an online auction over the internet. The auction-based market is thought to be more economically efficient than a traditional regulated market due to the interaction and easy access of different suppliers to the market. For developing a tool for power system planning, it is necessary to simulate a web-based auction in which human agents participate. However, participation of human agents in the market needs a lot of time and cost. A well-designed software agent can be a substitute to emulate the offer behavior of human agents. For a simulation, a limited number of different software agents participate, and each of the agents represent a firm that owns several generators. Thus, it is necessary to select agents for a simulation from all the software agents designed. Since it takes too much time to test all the possible permutations of all the possible software agents, a way is needed to classify the agents into a small number of groups based on the effect of the agents on the market outcomes. Several different types of strategy used by human agents have been observed in tests of markets using a uniform price auction. The most extreme strategies are to offer marginal cost and to speculate. For the sake of simplicity, earnings can be determined by the market clearing price and the quantity dispatched. A marginal cost offer agent wants to maximize the quantity dispatched by offering low and hopes that someone else sets a high price. A speculator wants to increase the market clearing price, and takes the risk of not getting as much capacity dispatched. It is fairly easy to model a marginal cost offer agent since it offers all the blocks at the marginal cost. On the other hand, there are many different types of speculators depending on the degree that they speculate.In this study, five standardized agents were designed for simulation and classification - four different types of speculators and a marginal cost offer agent. A human subject and a software agent competed against combinations of the standardized agents. Based on their performance (earnings in each period), the subjects and the agents were classified into the five different groups1.2. Electricity marketAgents develop auction rules for themselves based on the rules of the auction they are participating in and, in repeated auctions, based on the actions of their competitors. In the design considered here the electricity market was assumed to be a uniform price auction with an inelastic but time varying load demand. In this market an independent system operator (ISO) provides a load forecast and collects offers submitted by six participating agents. The ISO then clears the market and checks the security of the system.In every period, each agent is asked to submit a price and quantity. No price can exceed a reservation price meant to represent the price above which no load would be willing to pay for power. The offers submitted by all the agents are then ranked according to the offer price from lowest to highest. Then, the ISO dispatched blocks beginning with the lowest offer until actual demand (which is different than forecasted demand) is met. If two or more blocks were offered at the same price, the ISO randomly selected which block(s) to be dispatched. All the winning agents were paid according to a second price auction, meaning that winners were paid at the same price (uniform price auction). If the actual demand were larger than the capacity offered, ISO would recall short of capacity from the blocks withheld at the price of the last accepted offer. The agent whose block was recalled would be charged a recall cost. After clearing the market, ISO published the market clearing price and quantity dispatched to corresponding agents. Each agent received information only related to its own generator such as the dispatch quantity and price. One scenario was comprised of 200 periods.Six agents each had the same capacity with five blocks. Their generators had identical operating costs including fuel cost and standby cost as well as interest charges. For the sake of simplicity, startup costs were not taken into account. Based on its maximization algorithms, available history data and load forecast, each agent decided how many blocks to offer and the offer price of a block if offered. Exchange of information among agents was not allowed.1.3. Standardized agentsFive standardized agents consisting of one marginal cost offer agent and four speculators were designed to be used in a test bed whose purpose is to classify other software or human agents. That is, the thesis is that an agent with unknown behavior can be classified based on its play with known agent types. The marginal cost offer agent (MC) is an agent that offers all five blocks at marginal cost without any withholding. The four speculators had different degrees of speculation. In order to be a speculator, at least one block must be offered at a high price.It is crucial to an ability to implement a speculator to be able to determine which block or blocks are to be offered at a high price. For simplicity, any offer submitted at a high price was made at the same price regardless of the type of speculation. A fair share of the market was calculated based on the load forecast. The block in which the fair share quantity falls is termed the “fairshare block”. If this were the last block chosen for the unit by the auction, then it would be the units’ marginal block. Thus, the fair share calculation is just a means for trying to predict a unit’s marginal block a period ahead and any calculation that accomplishes that prediction is suitable for the purpose we have in mind. Since all the competitors in the market considered here have the same capacity, fairshare was calculated simply by dividing the load forecast by the number of market participants. If there were differences in the generating capacity being represented by an agent, the formula for a fair share is more complicate. Also, if some agents have a locational benefit over others, their fairshare should not be a simple。