下载文档原格式
各向异性完全匹配层边界条件下雷达散射截面的误差修正王向华;郑宏兴
【期刊名称】《天津职业技术师范大学学报》
【年(卷),期】2011(021)001
【摘要】为了提高时域有限差分法的计算精度,以二维金属圆柱散射为例,研究了在各向异性完全匹配层(UPML)吸收条件下,改变总场-散射场的输出及吸收边界时误差的变化规律.比较了UPML与二阶Mur吸收边界条件下的误差.同时发现,可采用迭代步数相差半个周期的双站雷达散射截面(RCS)的平均来减小误差,也可采用距离为半个波长的两个输出边界得到RCS的平均来减小误差.
【总页数】5页(P7-10,18)
【作者】王向华;郑宏兴
【作者单位】天津职业技术师范大学天线与微波技术研究所,天津,300222;天津职业技术师范大学天线与微波技术研究所,天津,300222
【正文语种】中文
【中图分类】TN957.51
【相关文献】
1.二阶Mur吸收边界条件下雷达散射截面的误差修正 [J], 王向华;郑宏兴;麻来宣
2.截断各向异性等离子体的修正各向异性完全匹配层吸收边界 [J], 杨利霞;王祎君;谢应涛;王刚
3.基于单轴各向异性完全匹配层边界条件的ADI-FDTD三维GPR全波场正演 [J], 冯德山;谢源
4.各向异性完全匹配层边界条件下雷达散射截面的误差修正 [J], 王向华;郑宏兴
5.各向异性完全匹配层边界条件下雷达散射截面的误差修正 [J], 王向华;郑宏兴;因版权原因，仅展示原文概要，查看原文内容请购买。
Linux MultiPath多路径部署方案目录一、环境摘要 (3)二、检查安装multipath (3)2.1检查是否已安装模块 (3)2.2加载模块 (3)2.3设为开机自动启动 (3)三、配置multipath (4)3.1创建配置脚本 (4)3.2确认配置结果 (8)3.3正常使用磁盘 (8)四、测试 (8)4.1负载均衡测试 (8)五、常用操作命令 (9)5.1启停mulitipath服务 (9)5.2删除现有路径 (9)5.3格式化路径（重新扫描） (9)5.4查看多路径 (9)5.5重载multipathd服务 (10)一、环境摘要Suse 11 sp1二、检查安装multipath2.1检查是否已安装模块Suse 11 sp2系统默认安装有以下模块# rpm -qa |grep mapperdevice-mapper-32bit-1.02.63-18.25.1device-mapper-1.02.63-18.25.1# rpm -qa |grep multimultipath-tools-0.4.9-0.60.1# rpm -qa |grep udevlibudev0-32bit-147-0.47.2libgudev-1_0-0-147-0.47.2udev-147-0.47.2libudev0-147-0.47.22.2加载模块modprobe dm-multipathmodprobe dm-round-robin2.3设为开机自动启动chkconfig --level 35 multipathd on三、配置multipath3.1创建配置脚本Multipath的配置文件是/etc/multipath.conf，将以下内容复制到文件中，并保存：以下脚本屏蔽了华为多路径软件，并且对多路径盘进行绑定，避免系统重启后设备名发生改变。
blacklist {wwid "3600605b0042372d01159327313feb8a2" devnode"^(ram|raw|loop|fd|md|dm-|sr|scd|st)[0-9]*" devnode "^hd[a-z]"device{vendor "up" #屏蔽华为多路径盘product "updisk"}}defaults {udev_dir /devpolling_interval 10#prio alua#rr_min_io 10user_friendly_names yesno_path_retry 10path_checker turpath_grouping_policy failover}devices {device {vendor "HP|COMPAQ"product "HSV1[01]1\(C\)COMPAQ|HSV2[01]0|HSV300|HSV4[05]0"path_grouping_policy group_by_priogetuid_callout "/lib/udev/scsi_id -g -u /dev/%n"path_checker turpath_selector "round-robin 0"prio aluarr_weight uniformfailback immediatehardware_handler "0"no_path_retry 18rr_min_io 100}}multipaths {multipath {wwid 3600050ccd64326f8579ee51180c744a8alias mpathapath_grouping_policy failoverfailback manualno_path_retry 10}multipath {wwid 3600050ccd44326f8579ee51180c744a8alias mpathbpath_grouping_policy failoverfailback manualno_path_retry 10}multipath {wwid 3600050ccdc4326f8579ee51180c744a8 alias mpathcpath_grouping_policy failover failback manualno_path_retry 10}multipath {wwid 3600050ccde4326f8579ee51180c744a8 alias mpathdpath_grouping_policy failover failback manualno_path_retry 10}multipath {wwid 3600050cce04326f8579ee51180c744a8 alias mpathepath_grouping_policy failover failback manualno_path_retry 10}}3.2确认配置结果1、启动服务：Service multipathd restart2、多路径聚合与查看多路径：Multipath –v2multipath -ll3、其他核查方式✓会在/dev/mapper/目录下多出类似mpatha、mpathb之类设备（如果指定了alias则以别名方式显示）✓用fdisk -l命令可以看到多路径软件创建的磁盘，如/dev/dm-[0-3]等3.3正常使用磁盘要使用多路径生成的磁盘直接操作/dev/mapper/目录下的磁盘就便可。
Red Hat Enterprise Linux 5DM MultipathDM Multipath Configuration and AdministrationDM MultipathRed Hat Enterprise Linux 5 DM MultipathDM Multipath Configuration and AdministrationEdition 3Copyright © 2010 Red Hat, Inc.The text of and illustrations in this document are licensed by Red Hat under a Creative Commons Attribution–Share Alike 3.0 Unported license ("CC-BY-SA"). An explanation of CC-BY-SA is available at /licenses/by-sa/3.0/. In accordance with CC-BY-SA, if you distribute this document or an adaptation of it, you must provide the URL for the original version.Red Hat, as the licensor of this document, waives the right to enforce, and agrees not to assert, Section 4d of CC-BY-SA to the fullest extent permitted by applicable law.Red Hat, Red Hat Enterprise Linux, the Shadowman logo, JBoss, MetaMatrix, Fedora, the Infinity Logo, and RHCE are trademarks of Red Hat, Inc., registered in the United States and other countries. Linux® is the registered trademark of Linus Torvalds in the United States and other countries.Java® is a registered trademark of Oracle and/or its affiliates.XFS® is a trademark of Silicon Graphics International Corp. or its subsidiaries in the United States and/or other countries.MySQL® is a registered trademark of MySQL AB in the United States, the European Union and other countries.All other trademarks are the property of their respective owners.1801 Varsity DriveRaleigh, NC 27606-2072 USAPhone: +1 919 754 3700Phone: 888 733 4281Fax: +1 919 754 3701This book provides information on using the Device-Mapper Multipath feature of Red Hat Enterprise Linux 5Preface v1. Audience (v)2. Related Documentation (v)3. Feedback (v)4. Document Conventions (vi)4.1. Typographic Conventions (vi)4.2. Pull-quote Conventions (vii)4.3. Notes and Warnings (viii)1. Device Mapper Multipathing 11.1. Overview of DM-Multipath (1)1.2. Storage Array Support (3)1.3. DM-Multipath Components (3)1.4. DM-Multipath Setup Overview (4)2. Multipath Devices 52.1. Multipath Device Identifiers (5)2.2. Consistent Multipath Device Names in a Cluster (5)2.3. Multipath Device Attributes (6)2.4. Multipath Devices in Logical Volumes (6)3. Setting Up DM-Multipath 73.1. Setting Up DM-Multipath (7)3.2. Ignoring Local Disks when Generating Multipath Devices (8)3.3. Adding Devices to the Multipathing Database (10)4. The DM-Multipath Configuration File 114.1. Configuration File Overview (11)4.2. Configuration File Blacklist (12)4.2.1. Blacklisting by WWID (12)4.2.2. Blacklisting By Device Name (12)4.2.3. Blacklisting By Device Type (13)4.2.4. Blacklist Exceptions (13)4.3. Configuration File Defaults (14)4.4. Multipaths Device Configuration Attributes (17)4.5. Configuration File Devices (18)5. DM-Multipath Administration and Troubleshooting 235.1. Resizing an Online Multipath Device (23)5.2. The Multipath Daemon (24)5.3. Issues with Large Number of LUNs (24)5.4. Issues with queue_if_no_path feature (24)5.5. Multipath Command Output (24)5.6. Multipath Queries with multipath Command (25)5.7. Multipath Command Options (26)5.8. Determining Device Mapper Entries with the dmsetup Command (26)5.9. Troubleshooting with the multipathd Interactive Console (26)A. Revision History 29 Index 31iiiivPrefaceThis book describes the Device Mapper Multipath (DM-Multipath) feature of Red Hat Enterprise Linux for the RHEL 5 release.1. AudienceThis book is intended to be used by system administrators managing systems running the Linux operating system. It requires familiarity with Red Hat Enterprise Linux.2. Related DocumentationFor more information about using Red Hat Enterprise Linux, refer to the following resources:•Red Hat Enterprise Linux Installation Guide — Provides information regarding installation of Red Hat Enterprise Linux 5.•Red Hat Enterprise Linux Deployment Guide — Provides information regarding the deployment, configuration and administration of Red Hat Enterprise Linux 5.For more information about Red Hat Cluster Suite for Red Hat Enterprise Linux 5, refer to the following resources:•Red Hat Cluster Suite Overview — Provides a high level overview of the Red Hat Cluster Suite.•Configuring and Managing a Red Hat Cluster — Provides information about installing, configuring and managing Red Hat Cluster components.•Logical Volume Manager Administration — Provides a description of the Logical Volume Manager (LVM), including information on running LVM in a clustered environment.•Global File System: Configuration and Administration — Provides information about installing, configuring, and maintaining Red Hat GFS (Red Hat Global File System).•Global File System 2: Configuration and Administration — Provides information about installing, configuring, and maintaining Red Hat GFS2 (Red Hat Global File System 2).•Using GNBD with Global File System — Provides an overview on using Global Network Block Device (GNBD) with Red Hat GFS.•Linux Virtual Server Administration — Provides information on configuring high-performance systems and services with the Linux Virtual Server (LVS).•Red Hat Cluster Suite Release Notes — Provides information about the current release of Red Hat Cluster Suite.Red Hat Cluster Suite documentation and other Red Hat documents are available in HTML,PDF, and RPM versions on the Red Hat Enterprise Linux Documentation CD and online at http:// /docs/.3. FeedbackIf you spot a typo, or if you have thought of a way to make this manual better, we would love tohear from you. Please submit a report in Bugzilla (/bugzilla/) against the component Documentation-cluster.vPrefaceBe sure to mention the manual's identifier:By mentioning this manual's identifier, we know exactly which version of the guide you have.If you have a suggestion for improving the documentation, try to be as specific as possible. If you have found an error, please include the section number and some of the surrounding text so we can find it easily.4. Document ConventionsThis manual uses several conventions to highlight certain words and phrases and draw attention to specific pieces of information.In PDF and paper editions, this manual uses typefaces drawn from the Liberation Fonts1 set. The Liberation Fonts set is also used in HTML editions if the set is installed on your system. If not, alternative but equivalent typefaces are displayed. Note: Red Hat Enterprise Linux 5 and later includes the Liberation Fonts set by default.4.1. Typographic ConventionsFour typographic conventions are used to call attention to specific words and phrases. These conventions, and the circumstances they apply to, are as follows.Mono-spaced BoldUsed to highlight system input, including shell commands, file names and paths. Also used to highlight keycaps and key combinations. For example:To see the contents of the file my_next_bestselling_novel in your currentworking directory, enter the cat my_next_bestselling_novel command at theshell prompt and press Enter to execute the command.The above includes a file name, a shell command and a keycap, all presented in mono-spaced bold and all distinguishable thanks to context.Key combinations can be distinguished from keycaps by the hyphen connecting each part of a key combination. For example:Press Enter to execute the command.Press Ctrl+Alt+F2 to switch to the first virtual terminal. Press Ctrl+Alt+F1 toreturn to your X-Windows session.The first paragraph highlights the particular keycap to press. The second highlights two key combinations (each a set of three keycaps with each set pressed simultaneously).If source code is discussed, class names, methods, functions, variable names and returned values mentioned within a paragraph will be presented as above, in mono-spaced bold. For example:1 https:///liberation-fonts/viPull-quote ConventionsviiFile-related classes include filesystem for file systems, file for files, and dir fordirectories. Each class has its own associated set of permissions.Proportional BoldThis denotes words or phrases encountered on a system, including application names; dialog box text;labeled buttons; check-box and radio button labels; menu titles and sub-menu titles. For example:Choose System → Preferences → Mouse from the main menu bar to launch MousePreferences . In the Buttons tab, click the Left-handed mouse check box and click Close to switch the primary mouse button from the left to the right (making the mousesuitable for use in the left hand).To insert a special character into a gedit file, choose Applications → Accessories→ Character Map from the main menu bar. Next, choose Search → Find… from theCharacter Map menu bar, type the name of the character in the Search field and click Next . The character you sought will be highlighted in the Character Table . Double-click this highlighted character to place it in the Text to copy field and then click theCopy button. Now switch back to your document and choose Edit → Paste from thegedit menu bar.The above text includes application names; system-wide menu names and items; application-specific menu names; and buttons and text found within a GUI interface, all presented in proportional bold and all distinguishable by context.Mono-spaced Bold Italic or Proportional Bold ItalicWhether mono-spaced bold or proportional bold, the addition of italics indicates replaceable orvariable text. Italics denotes text you do not input literally or displayed text that changes depending on circumstance. For example:To connect to a remote machine using ssh, type ssh username @ ata shell prompt. If the remote machine is and your username on thatmachine is john, type ssh john@ .The mount -o remount file-system command remounts the named filesystem. For example, to remount the /home file system, the command is mount -oremount /home .To see the version of a currently installed package, use the rpm -q packagecommand. It will return a result as follows: package-version-release .Note the words in bold italics above — username, , file-system, package, version and release. Each word is a placeholder, either for text you enter when issuing a command or for text displayed by the system.Aside from standard usage for presenting the title of a work, italics denotes the first use of a new and important term. For example:Publican is a DocBook publishing system.4.2. Pull-quote ConventionsTerminal output and source code listings are set off visually from the surrounding text.Output sent to a terminal is set in mono-spaced roman and presented thus:PrefaceviiiSource-code listings are also set in mono-spaced romanbut add syntax highlighting as follows:4.3. Notes and WarningsFinally, we use three visual styles to draw attention to information that might otherwise be overlooked.Chapter 1.Device Mapper MultipathingDevice Mapper Multipathing (DM-Multipath) allows you to configure multiple I/O paths between server nodes and storage arrays into a single device. These I/O paths are physical SAN connections that can include separate cables, switches, and controllers. Multipathing aggregates the I/O paths, creating a new device that consists of the aggregated paths.1.1. Overview of DM-MultipathDM-Multipath can be used to provide:•RedundancyDM-Multipath can provide failover in an active/passive configuration. In an active/passive configuration, only half the paths are used at any time for I/O. If any element of an I/O path (the cable, switch, or controller) fails, DM-Multipath switches to an alternate path.•Improved PerformanceDM-Multipath can be configured in active/active mode, where I/O is spread over the paths in a round-robin fashion. In some configurations, DM-Multipath can detect loading on the I/O paths and dynamically re-balance the load.Figure 1.1, “Active/Passive Multipath Configuration with One RAID Device” shows an active/passive configuration with two I/O paths from the server to a RAID device. There are 2 HBAs on the server, 2 SAN switches, and 2 RAID controllers.Figure 1.1. Active/Passive Multipath Configuration with One RAID DeviceIn this configuration, there is one I/O path that goes through hba1, SAN1, and controller 1 and a second I/O path that goes through hba2, SAN2, and controller2. There are many points of possible failure in this configuration:•HBA failure1Chapter 1. Device Mapper Multipathing 2•FC cable failure•SAN switch failure•Array controller port failureWith DM-Multipath configured, a failure at any of these points will cause DM-Multipath to switch to the alternate I/O path.Figure 1.2, “Active/Passive Multipath Configuration with Two RAID Devices” shows a more complex active/passive configuration with 2 HBAs on the server, 2 SAN switches, and 2 RAID devices with 2RAID controllers each.Figure 1.2. Active/Passive Multipath Configuration with Two RAID DevicesIn the example shown in Figure 1.2, “Active/Passive Multipath Configuration with Two RAID Devices”,there are two I/O paths to each RAID device (just as there are in the example shown in Figure 1.1,“Active/Passive Multipath Configuration with One RAID Device”). With DM-Multipath configured, a failure at any of the points of the I/O path to either of the RAID devices will cause DM-Multipath to switch to the alternate I/O path for that device.Figure 1.3, “Active/Active Multipath Configuration with One RAID Device” shows an active/active configuration with 2 HBAs on the server, 1 SAN switch, and 2 RAID controllers. There are four I/O paths from the server to a storage device:•hba1 to controller1•hba1 to controller2•hba2 to controller1•hba2 to controller2In this configuration, I/O can be spread among those four paths.Storage Array SupportFigure 1.3. Active/Active Multipath Configuration with One RAID Device1.2. Storage Array SupportBy default, DM-Multipath includes support for the most common storage arrays that support DM-Multipath. The supported devices can be found in the multipath.conf.defaults file. If your storage array supports DM-Multipath and is not configured by default in this file, you may need to add them to the DM-Multipath configuration file, multipath.conf. For information on the DM-Multipath configuration file, see Chapter 4, The DM-Multipath Configuration File.Some storage arrays require special handling of I/O errors and path switching. These require separate hardware handler kernel modules.1.3. DM-Multipath ComponentsTable 1.1, “DM-Multipath Components”. describes the components of DM-Multipath.Chapter 1. Device Mapper Multipathing1.4. DM-Multipath Setup OverviewDM-Multipath includes compiled-in default settings that are suitable for common multipath configurations. Setting up DM-multipath is often a simple procedure.The basic procedure for configuring your system with DM-Multipath is as follows:1.Install device-mapper-multipath rpm.2.Edit the multipath.conf configuration file:•comment out the default blacklist•change any of the existing defaults as needed•save the configuration file3.Start the multipath daemons.4.Create the multipath device with the multipath command.Detailed setup instructions for several example multipath configurations are provided in see Chapter 3, Setting Up DM-Multipath.Chapter 2.Multipath DevicesWithout DM-Multipath, each path from a server node to a storage controller is treated by the system as a separate device, even when the I/O path connects the same server node to the same storage controller. DM-Multipath provides a way of organizing the I/O paths logically, by creating a single multipath device on top of the underlying devices.2.1. Multipath Device IdentifiersEach multipath device has a World Wide Identifier (WWID), which is guaranteed to be globally unique and unchanging. By default, the name of a multipath device is set to its WWID. Alternately, you can set the user_friendly_names option in the multipath configuration file, which sets the alias to a node-unique name of the form mpath n.For example, a node with two HBAs attached to a storage controller with two ports via a single unzoned FC switch sees four devices: /dev/sda, /dev/sdb, dev/sdc, and /dev/sdd. DM-Multipath creates a single device with a unique WWID that reroutes I/O to those four underlying devices according to the multipath configuration. When the user_friendly_names configuration option is set to yes, the name of the multipath device is set to mpath n.When new devices are brought under the control of DM-Multipath, the new devices may be seen in three different places under the /dev directory: /dev/mapper/mpath n, /dev/mpath/mpath n, and /dev/dm-n.•The devices in /dev/mapper are created early in the boot process. Use these devices to access the multipathed devices, for example when creating logical volumes.•The devices in /dev/mpath are provided as a convenience so that all multipathed devices canbe seen in one directory. These devices are created by the udev device manager and may not be available on startup when the system needs to access them. Do not use these devices for creating logical volumes or filesystems.•Any devices of the form /dev/dm-n are for internal use only and should never be used.For information on the multipath configuration defaults, including the user_friendly_names configuration option, see Section 4.3, “Configuration File Defaults”.You can also set the name of a multipath device to a name of your choosing by using the alias option in the multipaths section of the multipath configuration file. For information on the multipaths section of the multipath configuration file, see Section 4.4, “Multipaths Device Configuration Attributes”.2.2. Consistent Multipath Device Names in a ClusterWhen the user_friendly_names configuration option is set to yes, the name of the multipath device is unique to a node, but it is not guaranteed to be the same on all nodes using the multipath device. This should not cause any difficulties if you use LVM to create logical devices from the multipath device, but if you require that your multipath device names be consistent in every node in the cluster you perform one of the following procedures:•Use the alias option in the multipaths section of the multipath configuration file to set the name of the multipath device. The alias for the multipath device is consistent across all the nodes ina cluster. For information on the multipaths section of the multipath configuration file, see see Section 4.4, “Multipaths Device Configuration Attributes”.Chapter 2. Multipath Devices•If you want the system-defined user-friendly names to be consistent across all nodes in the cluster, set up all of the multipath devices on one machine. Then copy the /var/lib/multipath/ bindings file from that machine to all the other machines in the cluster.2.3. Multipath Device AttributesIn addition to the user_friendly_names and alias options, a multipath device has numerous attributes. You can modify these attributes for a specific multipath device by creating an entry forthat device in the multipaths section of the multipath configuration file. For information on the multipaths section of the multipath configuration file, see see Section 4.4, “Multipaths Device Configuration Attributes”.2.4. Multipath Devices in Logical VolumesAfter creating multipath devices, you can use the multipath device names just as you would use a physical device name when creating an LVM physical volume. For example, if /dev/mapper/mpath0 is the name of a multipath device, the following command will mark /dev/mapper/mpath0 as a physical volume.You can use the resulting LVM physical device when you create an LVM volume group just as you would use any other LVM physical device.When you create an LVM logical volume that uses active/passive multipath arrays as the underlying physical devices, you should include filters in the lvm.conf to exclude the disks that underlie the multipath devices. This is because if the array automatically changes the active path to the passive path when it receives I/O, multipath will failover and failback whenever LVM scans the passive path if these devices are not filtered. For active/passive arrays that require a command to make the passive path active, LVM prints a warning message when this occurs.To filter all SCSI devices in the LVM configuration file (lvm.conf), include the following filter in the devices section of the file.Chapter 3.Setting Up DM-MultipathThis chapter provides step-by-step example procedures for configuring DM-Multipath. It includes the following procedures:•Basic DM-Multipath setup•Ignoring local disks•Adding more devices to the configuration file3.1. Setting Up DM-MultipathBefore setting up DM-Multipath on your system, ensure that your system has been updated and includes the device-mapper-multipath package.Use the following procedure to set up DM-Multipath for a basic failover configuration.1.Edit the /etc/multipath.conf file by commenting out the following lines at the top of the file.This section of the configuration file, in its initial state, blacklists all devices. You must comment it out to enable multipathing.After commenting out those lines, this section appears as follows.2.The default settings for DM-Multipath are compiled in to the system and do not need to beexplicitly set in the /etc/multipath.conf file.The default value of path_grouping_policy is set to failover, so in this example you do not need to change the default value. For information on changing the values in the configuration file to something other than the defaults, see Chapter 4, The DM-Multipath Configuration File.The initial defaults section of the configuration file configures your system that the names of the multipath devices are of the form mpath n; without this setting, the names of the multipath devices would be aliased to the WWID of the device.3.Save the configuration file and exit the editor.4.Execute the following commands:Chapter 3. Setting Up DM-MultipathThe multipath -v2 command prints out multipathed paths that show which devices aremultipathed. If the command does not print anything out, ensure that all SAN connections are set up properly and the system is multipathed.For further information on the multipath command output, see Section 5.5, “MultipathCommand Output”.5.Execute the following command to ensure sure that the multipath daemon starts on bootup:Since the value of user_friendly_name is set to yes in the configuration file the multipath devices will be created as /dev/mapper/mpath n. For information on setting the name of the device to an alias of your choosing, see Chapter 4, The DM-Multipath Configuration File.3.2. Ignoring Local Disks when Generating Multipath DevicesSome machines have local SCSI cards for their internal disks. DM-Multipath is not recommended for these devices. The following procedure shows how to modify the multipath configuration file to ignore the local disks when configuring multipath.1.Determine which disks are the internal disks and mark them as the ones to blacklist.In this example, /dev/sda is the internal disk. Note that as originally configured in the default multipath configuration file, executing the multipath -v2 shows the local disk, /dev/sda, in the multipath map.For further information on the multipath command output, see Section 5.5, “MultipathCommand Output”.Ignoring Local Disks when Generating Multipath Devices2.In order to prevent the device mapper from mapping /dev/sda in its multipath maps, edit theblacklist section of the /etc/multipath.conf file to include this device. Although you could blacklist the sda device using a devnode type, that would not be safe procedure since /dev/sda is not guaranteed to be the same on reboot. To blacklist individual devices, you can blacklist using the WWID of that device.Note that in the output to the multipath -v2 command, the WWID of the /dev/sda device is SIBM-ESXSST336732LC____F3ET0EP0Q000072428BX1. To blacklist this device, include the following in the /etc/multipath.conf file.3.After you have updated the /etc/multipath.conf file, you must manually tell themultipathd daemon to reload the file. The following command reloads the updated /etc/multipath.conf file.4.Run the following commands:The local disk or disks should no longer be listed in the new multipath maps, as shown in the following example.Chapter 3. Setting Up DM-Multipath3.3. Adding Devices to the Multipathing DatabaseBy default, DM-Multipath includes support for the most common storage arrays that support DM-Multipath. The default configuration values, including supported devices, can be found in the multipath.conf.defaults file.If you need to add a storage device that is not supported by default as a known multipath device, edit the /etc/multipath.conf file and insert the appropriate device information.For example, to add information about the HP Open-V series the entry looks like this:For more information on the devices section of the configuration file, see Section 4.5, “Configuration File Devices”.Chapter 4.The DM-Multipath Configuration FileBy default, DM-Multipath provides configuration values for the most common uses of multipathing.In addition, DM-Multipath includes support for the most common storage arrays that support DM-Multipath. The default configuration values and the supported devices can be found in the /usr/ share/doc/device-mapper-multipath-0.4.7/multipath.conf.defaults file.You can override the default configuration values for DM-Multipath by editing the /etc/ multipath.conf configuration file. If necessary, you can also add a storage array that is not supported by default to the configuration file. This chapter provides information on parsing and modifying the multipath.conf file. It contains sections on the following topics:•Configuration file overview•Configuration file blacklist•Configuration file defaults•Configuration file multipaths•Configuration file devicesIn the multipath configuration file, you need to specify only the sections that you needfor your configuration, or that you wish to change from the default values specified in the multipath.conf.defaults file. If there are sections of the file that are not relevant to your environment or for which you do not need to override the default values, you can leave them commented out, as they are in the initial file.The configuration file allows regular expression description syntax.An annotated version of the configuration file can be found in /usr/share/doc/device-mapper-multipathd-0.4.7/multipath.conf.annotated.4.1. Configuration File OverviewThe multipath configuration file is divided into the following sections:blacklistListing of specific devices that will not be considered for multipath. By default all devices are blacklisted. Usually the default blacklist section is commented out.blacklist_exceptionsListing of multipath candidates that would otherwise be blacklisted according to the parameters of the blacklist section.defaultsGeneral default settings for DM-Multipath.multipathsSettings for the characteristics of individual multipath devices. These values overwrite what is specified in the defaults and devices sections of the configuration file.devicesSettings for the individual storage controllers. These values overwrite what is specified in the defaults section of the configuration file. If you are using a storage array that is not supported by default, you may need to create a devices subsection for your array.。
Multisin学习一、multisim菜单栏multisim10有12个主菜单，如图1.2.2所示，菜单中提供了本软件几乎所有的功能命令。
1. File（文件）菜单File（文件）菜单提供19个文件操作命令，如打开、保存和打印等，File菜单中的命令及功能如下：New：建立一个新文件。
Open：打开一个己存在的*．msm10、*．msm9、*．msm8、*．msm7、*．ewb或*．utsch等格式的文件。
Close：关闭当前电路工作区内的文件。
Close All：关闭电路工作区内的所有文件。
Save：将电路工作区内的文件以*．msm10的格式存盘。
Save as：将电路工作区内的文件另存为一个文件，仍为*．msm10格式。
Save All：将电路工作区内所有的文件以*．msm10的格式存盘。
New Project：建立新的项目（仅在专业版中出现，教育版中无此功能）。
Open Project：打开原有的项目（仅在专业版中出现，教育版中无此功能）。
Save Project：保存当前的项目（仅在专业版中出现，教育版中无此功能）。
Close Project：关闭当前的项目（仅在专业版中出现，教育版中无此功能）。
Version Control：版本控制（仅在专业版中出现，教育版中无此功能）。
Print：打印电路工作区内的电原理图。
Print Preview：打印预览。
Print Options: 包括Print Setup（打印设置）和Print Instruments（打印电路工作区内的仪表）命令。
Recent Files：选择打开最近打开过的文件。
Recent Projects：选择打开最近打开过的项目。
Exit：退出。
2. Edit（编辑）菜单Edit（编辑）菜单在电路绘制过程中，提供对电路和元件进行剪切、粘贴、旋转等操作命令，共21个命令，Edit菜单中的命令及功能如下：Undo：取消前一次操作。
超宽带（Ultra-Wide Band，简称UWB）全球定位系统（Global Positioning System，简称GPS）脉冲无线电（Impulse Radio，简称IR）联邦通信委员会（Federal Communications Commision，简称FCC）有效各向同性发射功率（Effective Isotropic Radiated Power，简称EIRP）基于位置服务（Location Based Service，简称LBS）射频识别（Radio Frequency Identification ,简称RFID）全球导航卫星系统(Global Navigation Satellite System，简称GNSS)视距(Line of Sight，简称LOS)三角定位系统（Triangulation-based Systems）三边定位系统（Trilateration-based Systems）到达时间（Time of Arrival，简称TOA）到达时间差（Time Difference of Arrival，简称TDOA）到达角度（Angle of Arrival，简称AOA）接收信号强度（Received Signal Strength，简称RSS）最大似然（Maximum Likelihood，简称ML）资产位置精度（Precision Asset Location，简称PAL）均方根误差（Root Mean Square Error，简称RMSE）非视距（Non Line of Sight，简称NLOS ）加性白噪声(Additive White Gaussian Noise，简称AWGN)高斯脉冲（Gaussian Pulse）脉冲幅度调制（Pulse Amplitude Modulation，简称PAM）脉冲位置调制（Pulse Position Modulation，简称PPM）脉冲波形调制（Pulse Shape Modulation，简称PSM）开关键控（On-Off Keying，简称OOK）二进制相移键控（Binary Phase Shift Keying，简称BPSK）跳时扩频通信系统（Time Hopping Spread Spectrum Communication Systems，简称TH-SS）最小二乘（Least-Squares Estimation，简称LSE）GPS辅助（Assisted-GPS，简称A-GPS）红外线技术（Infared）蓝牙（Bluetooth）无线局域网（Wireless Local Aera Ntework，WLAN）无线传感器网络（Wireless Sensor Network，简称WSN）无线个域网（Wireless Personal Aera Network，简称WPAN）直达径（Direct Path，简称DP）最强径（Strongest Path，简称SP）有明显直达经（Dominant Direct Path，简称DDP）无明显直达经（Nondominant Direct Path，简称NDDP）无直达经（Undetected Direct Path，简称UDP）广义极大似然（Generalized Maximum Likeliwood，简称GML）直接序列码分多址（Direct Squence Code Division Multiple Access，简称DS-CDMA）多频带正交频分复用联盟（Multi-Band Orthogonal Frequency Division Multiplexing Alliance，简称MBOA）信噪比（Signal-to-Noise Ratio，简称SNR）功率谱密度（Power Spectrum Density，简称PSD）能量谱密度（Energy Spectrum Density，简称ESD）无线广域网（Wireless Wide Aera Network，简称WWAN）无线城域网（Wireless Metropolitan Aera Network，简称WMAN）无线多媒体广播网（Wireless Multimedia Broadcast Network，简称WMBN）码分多址（Code Division Multiple Access，简称CDMA）移动广带无线接入（Mobile Broadband Wireless Access，简称MBWA）信道容量（Channel capacity）介质访问控制（Mediun Access Contril，简称MAC）最小二乘（Least Square，简称LS）克拉美罗下界（Cramer-Rao Lower Bound，简称CRLB）均方误差（Mean Square Error，简称MSE）均方根误差（Root Mean Square Error，简称RMSE）圆误差概率（Circular Error Probability，简称CEP）球误差概率（Spheric Error Probability，简称SEP）累计分布函数（Cumulative Distribution Function，简称CDF）几何精度因子（Geometric Dilution of Precision，简称GDOP）相对定位误差（Relative Position Error，简称RPE）加权最小二乘法（Weighted Least Squares，简称WLS）最大概率算法（High Probability Algorithm，简称HPA）平均值算法（Mean Value Algorithm，简称MVA）单次测量算法（One Time Algorithm，简称OTA）空间信号预测（Spatial Signal Prediction，简称SSP）匹配滤波（Matching Filtering，简称MF）能量探测（Energy Detector，简称ED）最大能量法（Maximum Energy Selection，简称MES）门限（Threshold Crossing，简称TC）固定门限法（Fixed Threshold，简称FT）平均绝对误差（Mean Absolute Error，简称MAE）均方差（Mean Square Error，简称MSE）标准差（Standard Deviation）最近邻法（Nearest Neighborhood，简称NN）K近邻法（K Nearest Neighborhood，简称KNN）信噪比（Signal to Noise Ratio，简称SNR）人工神经网络（Artificial Netural Networks，简称ANN）附加时延（Excess Delay）总多经增益（Total Multipath Gain）均方根时延扩展（Root Mean Square Delay Spread ，简称RMSDS）BP神经网络（Back Propagation）双程测距（Two Way Ranging，简称TWR）单程测距（One Way Ranging，简称OWR）卡尔曼滤波器（Kalman Filter，简称KF）扩展卡尔曼滤波器（Extend Kalman Filter，简称EKF）最大最小值比（Maximum to minimum energy sample ratio，简称MMR）门限比较算法（Threshold comparision，简称TC）误差累计分布函数（Cumulative Distribution Function，简称CDF）LNA（Low Noise Amplifier）低噪声放大器BPF（band-pass filter）带通滤波器PDF 概率分布函数（Probability Distribution Function）。
multipath.conf 参数传递dm
在multipath.conf文件中，可以使用参数来配置设备映射（dm）的相关设置。
以下是一些常用的参数：
1. user_friendly_names：设置为yes时，设备映射的名字会更加友好，易于识别，默认为no。
2. find_multipaths：设置为yes时，启用自动发现新的多路径设备，默认为yes。
3. verbose：设置为yes时，显示详细的多路径设备和路径的信息，默认为no。
4. selector：定义了要使用的多路径选择器。
可以选择的值有"round-robin"（轮询），"queue-length"（队列长度）或者其他自定义的选择器。
5. path_grouping_policy：定义了如何对路径进行分组，默认为failover。
其他可选的值有multipath（多路径），group_by_serial（按磁盘序列号分组）等。
6. failback：定义了故障回复方式，默认为immediate，即尽快切换回来。
其他可选的值有manual（手动切换）等。
7. rr_min_io：定义了在多路径选择器中，给每个路径分配的最小IO请求数量。
8. rr_weight：定义了在多路径选择器中，每个路径的权重。
9. rr_min_io_rq：定义了在多路径选择器中，分配给每个IO请求的最小路径数。
这些参数可以在multipath.conf文件的相应部分进行配置，以传递给dm设备管理程序。
Red Flag Asianux3上使用DMMultipath工具复合多条路径IBM存储v3500使用双控制器时，每台主机到存储之间有2对光纤跳线连接，主机映射完成后，即可在相应的服务器主机上看到该存储。
通过fdisk命令查看硬盘可以发现存储V3500上映射过来的卷，但是因为存在冗余的路径，所以每个卷看到了两个同样大小的存储空间。
如下面红色区域显示：[root@eCardDB02~]#fdisk-lDisk/dev/sda:299.4GB,299439751168bytes255heads,63sectors/track,36404cylindersUnits=cylinders of16065*512=8225280bytesDevice Boot Start End Blocks Id System/dev/sda1*11310439183Linux/dev/sda2143232525954614083Linux/dev/sda3323263640432764567+82Linux swap/SolarisWARNING:GPT(GUID Partition Table)detected on'/dev/sdb'!The util fdisk doesn't support e GNU Parted.WARNING:The size of this disk is6.0TB(5991479377920bytes).DOS partition table format can not be used on drives for volumeslarger than2.2TB(2199023255040bytes).Use parted(1)and GUIDpartition table format(GPT).Disk/dev/sdb:5991.4GB,5991479377920bytes255heads,63sectors/track,728422cylindersUnits=cylinders of16065*512=8225280bytesDevice Boot Start End Blocks Id System/dev/sdb112673502147483647+ee EFI GPTWARNING:GPT(GUID Partition Table)detected on'/dev/sdc'!The util fdisk doesn't support e GNU Parted.WARNING:The size of this disk is6.0TB(5991479377920bytes).DOS partition table format can not be used on drives for volumeslarger than2.2TB(2199023255040bytes).Use parted(1)and GUIDpartition table format(GPT).Disk/dev/sdc:5991.4GB,5991479377920bytes255heads,63sectors/track,728422cylindersUnits=cylinders of16065*512=8225280bytesDevice Boot Start End Blocks Id System/dev/sdc112673502147483647+ee EFI GPTWARNING:GPT(GUID Partition Table)detected on'/dev/dm-0'!The util fdisk doesn't support e GNU Parted.WARNING:The size of this disk is6.0TB(5991479377920bytes).DOS partition table format can not be used on drives for volumeslarger than2.2TB(2199023255040bytes).Use parted(1)and GUIDpartition table format(GPT).如上面所示看到有2个路径sdb和sdc，而且容量与存储划分的卷一致，而且2个路径不能用来分区和挂载。
Multipath Selection in Multi-radio Mesh Networks Irfan Sheriff Elizabeth Belding-RoyerDepartment of Computer ScienceUniversity of California,Santa Barbara{isheriff,ebelding}@AbstractResearch has shown that multi-radio multi-channel mesh networks provide significant capacity gains over single-radio mesh networks[10,20,21].Traditional single path routing can lead to poor utilization of the available channels in these networks.Opportunistic multipath routing can better exploit the available chan-nel diversity in a multi-radio network.The goal of this paper is to select multiple paths that,when used con-currently,provide high end-to-end throughput.To this end,we present a metric for multipath selection in multi-radio networks.We evaluate the metric through simu-lations in Qualnet and show that intelligent multipath routing significantly outperforms single path routing in multi-radio mesh networks.1.IntroductionMultihop wireless networks have emerged as an eco-nomical solution for providing last mile broadband ac-cess to the munity mesh networks,as they are popularly known,consist primarily of static wire-less mesh routers that together form a multihop network. With the deployment of a few wired gateways,Inter-net access is provided throughout the network.How-ever,due to the presence of a shared medium,the avail-able bandwidth decreases considerably as the number of hops from the gateway increases.Multi-radio net-works[10,20,21]that operate on multiple orthogonal frequency channels have been proposed as a solution to improve the capacity of mesh networks.We focus on these networks in this paper.Additional capacity is gained in multi-radio networks through simultaneous data transmission on multiple or-thogonal frequency channels.IEEE802.11a supports12 orthogonal channels and IEEE802.11b/g supports3or-thogonal channels.With the use of multiple radios tuned to different channels,a node can communicate with mul-tiple neighbors simultaneously.Nodes within the in-terference range of each other can avoid interference through communication over different channels.The ca-pacity of the network depends on the number of orthog-onal channels supported by a radio and the number of radios per node.When the number of radios is smaller than the available orthogonal channels,the capacity de-pends on the channel assignment in the network.Mul-tiple channel assignment techniques[10,20,21]have been proposed to achieve a capacity close to the optimal capacity of the network.In order to utilize the bandwidth available on multi-ple channels in a multi-radio network,the WCETT[7] metric selects high throughput,channel diverse paths. However,a single path might not be able to utilize the bandwidth available on all the channels in the neighbor-hood.This can happen when the channel diversity on a single path does not ensure the utilization of all available channels or when the average path length is less than the number of available channels.Thus there is a need for the use of multiple paths to exploit the bandwidth available on the additional channels on a opportunistic basis.We use the term‘opportunistic’because the gain depends on the channel assignment and the availability of additional paths from the source node.Multiple paths have been shown to be useful in single-radio networks to provide error resilience and load distribution[15,16]. The spatial diversity and data redundancy provided by multiple paths can be exploited for better end-to-end re-liability and traffic distribution.A number of multipath routing protocol solutions [9,11,12,17,22]have been proposed for ad hoc net-works that discover multiple paths between a source and destination.In each of these protocols,the term mul-tipath refers to the discovery of multiple paths that are actually used as backup paths.In a sense,then,these backup paths are actually alternate paths that are utilized only when the primary path fails.In this paper,the term multipath refers to the use of multiple concurrent paths between two nodes to increase the effective throughput. The goal of each previously proposed solution is to alle-viate the problem of link breaks in the network by com-puting and caching alternate paths obtained during the route discovery process.In these protocols,link failureon the primary path,on which the data transmission is taking place,causes the source to switch to an alternate path instead of re-initiating the route discovery process.A new route discovery is initiated only when all the pre-computed paths break.In this paper,we focus on exploiting channel diver-sity through multiple paths.Specifically,the contri-bution of this paper is a channel-aware routing metric for selection of multiple paths in multi-radio networks. We provide a comprehensive evaluation of the proposed metric through simulations in Qualnet and demonstrate that intelligent multipath selection can significantly out-perform single path routing in multi-radio networks.To the best of our knowledge,this is thefirst work on mul-tiple route selection for multi-radio networks.The rest of the paper is organized as follows.First, we provide the background in Section2.Section3 summarizes the goals and objectives.We then present the design of the path selection metric in Section4. Next,we evaluate the performance of the selection met-ric through simulations on Qualnet in Section5.The related work is presented in Section6.Finally,we sum-marize and conclude in Section7.2.BackgroundWefirst introduce a few terminologies used in the pa-per.We use‘multipath’to refer to a set of two or more paths,and use the terms‘multipath’and‘multipath com-bination’interchangeably.A‘Load ratio’of(x:y) on a multipath combination of paths I and II means that for every x packets on path I,y packets are transmitted on path II.‘Load ratio’and‘load distribution ratio’are used interchangeably.‘Maximize’and‘minimize’do not meanfinding the highest and lowest possible values, but refer to increasing as high as possible and decreasing as low as possible,respectively.We next introduce the link metric followed by the single path routing metric proposed for multi-radio net-works and then examine a simple network to motivate the need for multiple paths in multi-radio networks.2.1.Expected Transmission TimeThe ETT metric[7]represents the expected transmis-sion time of a packet on a link.ETT takes both the link bandwidth and the loss rate into consideration.If S denotes the packet length,B denotes the bandwidth of the link,and ETX[6]represents the expected number of transmissions,then the ETT is computed as follows:ETT=ETX∗S/B(1) The above calculation ignores the IEEE802.11backoff time.As shown in[7],the gain in accuracy taking theS DA30 msB15 ms30 ms15 msch 1ch 1ch 6ch 11Figure1.A two-radio IEEE802.11gnetwork.backoff into consideration is minimal.Hence we use Equation1to estimate the expected transmission time of a packet on a link.The ETT on a link closely represents the amount of time spent by a transmitted packet in the medium.2.2.Weighted Cumulative ETTDraves et al.[7]proposed the Weighted Cumulative ETT(WCETT)metric for route selection in multi-radio mesh networks.A weighted average of the bottleneck channel transmission time and the aggregate delay of all the links reflects the goodness in terms of the achievable throughput on a path.The channel with the highest ag-gregate ETT is the bottleneck channel for a path.The WCETT for a path is given by:WCETT=η∗BotETT+(1−η)∗AggrETT(2)where“BotETT”is the aggregate transmission time on the bottleneck channel and“AggrETT”is the aggregate transmission time on all the links.ηis a value between0 and1,and reflects the extent to which channel diversity and end-to-end characteristics are considered for path selection.2.3.A Motivating ExampleFigure1shows a simple two-radio network where each node is equipped with two IEEE802.11g radios. The numeric value on each link represents the ETT in milliseconds and the label“ch X”on a link denotes that channel X is used for communication.There are two paths available from S to D:path I with nodes S,A and D,and path II with nodes S,B and D.The ETTs of the links on path I are15ms on channel1and30ms on channel6.Channel6is thus the bottleneck channel thatlimits the throughput.Similarly,channel11,with an ETT of30ms,is the bottleneck channel on path II.Suppose we use path I for routing packets from S to D.The maximum achievable throughput is limited by the bottleneck ETT of30ms on the channel6.The achievable bottleneck throughput1is thus one packet ev-ery30ms.At this rate,channel1,with an ETT of15ms, is only about50%utilized while channel6is fully uti-lized.Further,channel11remains unused.On path II, the achievable bottleneck throughput is one packet ev-ery30ms on channel11.In this case,channel1is50% utilized at the bottleneck throughput rate,and channel6 remains unutilized.If the traffic is distributed equally on the two paths, the channel utilization is as follows:For every two pack-ets transmitted by the sender,a total of30ms time is spent on channels1,6and11since the packets can be transmitted concurrently on the paths.Thus the achiev-able bottleneck throughput is2packets every30ms or one packet every15ms.That is a100%improve-ment in the achievable bottleneck channel throughput of the multipath combination over either of the individual paths.Hence we expect a significant gain in the end-to-end throughput as well.Of course,throughput gain can vary since the end-to-end throughput is affected by other factors such as the hop count and the end-to-end characteristics[7].2.4.Traffic PartitioningMultipath routing necessitates traffic partitioning. Traffic distribution can be accomplished in multiple ways.One technique is to distribute multiple concurrent flows from the source to the gateway node on multiple paths.Theflows could belong to data or multimedia ap-plications.A single multimedia application such as a video conference or streaming video is typically divided into multipleflows.If a layered scheme of multimedia streaming is used at the source[23],a single multime-dia stream is divided into one base layer and several en-hancement layers.Another technique,multiple descrip-tion coding,splits a multimedia stream into multiple lay-ers of equal importance.Another method to partition traffic is to transmit packets from the sameflow on different paths.The drawback of this scheme is that it leads to out of order packet reception at the destination and can thus nega-tively impact the performance of higher layer protocols such as TCP.Real time applications,however,have a re-sequencing buffer at the receiver and can tolerate packet reordering within certain end-to-end delay limits.let SC(P)represent the set of channels on path P.We derive the goodness of the path combination(P a,P b)in two parts:Inter-path interference index(λ):The inter-path interference index(λ)accounts for the interference among the common channel links on the two paths. Individually,the throughput on the paths is limited by the respective bottleneck channels.When used together, the interference between the links on the two paths can limit the joint throughput achievable.We show that the bottleneck channel depends on the load distribution over the multipath combination,and compute the load ratio at which the average per packet bottleneck channel transmission time is minimized.Consider a path P k with m k channels.Define X kj as follows:X kj=Hop i is on channel jET T ki j∈SC(P k)(3) X kj represents the total per packet transmission time onchannel j and max1≤i≤mk X ki gives the transmissiontime on the bottleneck channel of the path.We nowfind the bottleneck channel transmission time for a multipath combination.Consider a traffic partition with a load ra-tio of(x:y)on paths P a and P b.For a channel j,x packets on path P a and y packets on path P b consume a channel transmission time of x∗X aj+y∗Y bj.The average per packet consumption time on channel j in SC(P a)∪SC(P b)can thus be written as:Y j(x,y)=x∗X aj+y∗X bjx+y as given byEquation4.This is because,for every x+y packets, only x packets consume the bandwidth on channel j.It is clear from Equation4that the traffic load ratio(x:y)determines the packet consumption time on the channels.The channel with the high-est packet consumption time acts as the bottleneck. Hence the bottleneck channel on a multipath can change with the load distribution ratio.If m ab represents the total number of channels on the path combina-tion(P a,P b),max1≤i≤mab Y i(x,y)gives the bottleneckchannel transmission time at the load distribution ratio (x:y).We thus defineλ(x,y)as follows:λ(x,y)=max1≤j≤m ab Y j(x,y)(5)Independent path quality index(γ):The independentpath quality index(γ)accounts for the end-to-end char-acteristics of the two paths.The WCETT metric givesthe path quality of a single path in a multi-radio net-work.The WCETT for a path P k with m k channels andn hops is given by:W CET T(P k)=η∗max1≤i≤m kX ki+(1−η)∗ni=1ET T ki(6)The higher the WCETT is on a path,the lower is the ex-pected throughput.To account for the end-to-end char-acteristics of both the paths,a weighted average of theWCETT on the paths is computed.We defineγ(x,y)asfollows:γ(x,y)=x∗W CET T(P a)+y∗W CET T(P b)I,with fewer hops,has lower WCETT and is expected to provide higher throughput.With no interference be-tween the paths,path I should share a higher proportion of the traffic load for good end-to-end performance of the multipath combination.It might seem that the individual path characteristics could be used to decide the load ratio on the paths.How-ever,this could be undesirable when there is channel in-terference between the paths.Consider the same topol-ogy in the above paragraph with modifications.Let path I consist of a single hop on channel34at an ETT of15 ms and let path II consist of three hops that operate on channels44and46at an ETT of10ms,and channel 34at an ETT of5ms.The two paths interfere on the common channel34.The WCETT of path II is higher than path I.However,the ETT on channel34on path I is higher than any other channel.If WCETT is used to decide on the traffic distribution,the slowest link of the multipath is loaded with more traffic,potentially re-sulting in under-utilization of the other channels on the multipath.When there is no channel interference between the paths,we expectλ(x,y)to have little impact on the end-to-end throughput of a multipath.Hence we select the load ratio in the inverse ratio of the WCETT of the paths when there is no common channel between the paths.When there is channel interference between the paths,we partition the traffic so as to minimize the inter-path interference index(λ).CAM:We define the Channel Aware Multipath (CAM)metric for selection of multiple paths as a combination ofλandγ:CAM=β∗λ+(1−β)∗γ(8) CAM accounts for both channel diversity between the paths and the end-to-end characteristics of the individ-ual paths.A low value of CAM corresponds to high throughput and vice versa.Ifβis chosen to be one,the metric only takes channel diversity between the paths into consideration.Ifβis zero,only the end-to-end char-acteristics of the paths determine the multipath selection.A value between zero and one forβaccounts for both.As an example,consider a3-radio network with four orthogonal channels.We construct multipaths1and 2from node S to node D as shown in Figure2.The channel and the ETT on the links are shown.Since there is a common channel between the paths on both the multipaths,the load ratio is selected to minimize the inter-path interference index(λ)on the multipaths.We plot the variation of the per packet consumption on each channel with the load distribution ratio.As given by Equation4,the per packet channel consumption time Y i(x,y)on channel i for a load ratio(x:y)can be writtenas:S D(a)Multipath1.DS(b)Multipath2.Figure 2.Two possible sets of multiplepaths available from node S to node D.Y1(x,y)=30xx+yand Y3(x,y)=20yx+y,Y2(x,y)=13.5x+10yx+yand Y4=10x+10y2ηcontrols the degree of importance of channel diversity over path length on a single path.Load ratio component x with y=1λ(x ,y )(a)Load fixed on path II.(b)Load fixed on path I.Figure 3.The variation of the inter-path interference index with the load distribution ratio.4.2.Interface Disjoint PathsIEEE 802.11uses exponential backoff upon packet failure before a retransmission takes place.Hence the channel utilization of a single interface reduces with an increase in the packet error rate.Multiple instances of 802.11could thus operate on the same channel and po-tentially provide better channel utilization,and thereby provide a gain in throughput.The analogy is similar to the gain obtained with multiple TCP connections [18].If the throughput with multiple interfaces on a common channel is significantly higher than the single interface throughput,the path selection metric should explicitly account for the degree of interface disjoint-ness between paths.We show in the Appendix that the additional com-plexity to evaluate the gain obtained with interface dis-joint paths does not justify the gain in throughput.Hence,CAM does not differentiate paths with a com-mon interface from interface disjoint paths.However,interface disjoint paths could provide better spatial di-versity than common interface paths.Quantifying the trade-off between the spatial and channel diversity gain is not clear and we plan to explore this in future.4.3.Revisiting Link InterferenceIn our analysis,we have assumed that all the links on a multipath interfere.This is a pessimistic assumption and does not hold true for longer spatially diverse paths.This assumption can be relaxed by building the inter-ference graph for the mesh network.The interference graph G =(V,E )consists of a set of vertices connected by edges.The vertices represent the links in the network and an edge between two vertices indicates that the two links representing the vertices interfere with each other.Padhye et al.[19]proposed an 0(n 2)algorithm for mea-suring the all-pair link interference in a n node network.We define X pkj as follows:X pkj = Hop i on channel j interferes pET T ki (9)where the link p operates on channel j and X pkj repre-sents the aggregate ETT of links interfering with linkp .X pkj represents the total transmission time on all the links interfering with link p .The interfering links can be obtained from the interference graph G.The maximumof X p kj on all the links operating on channel j represents the channel consumption time X kj on a channel j .X kj can thus be given by:X kj =maxlink p on channel jX p kj(10)We can compute the inter-path interference index (λ)by taking the per link interference into consideration and computing the maximum Y j (x,y )over all links.In this paper,we compute λand γby assuming all the links to be in the interfering range of each other.5.Implementation and EvaluationIn this section,we evaluate the effectiveness of CAM in selecting high throughput multiple paths and we quan-tify the gain in throughput with multiple paths in multi-channel networks.The Qualnet simulator is used for the perfor-mance evaluation.The Optimized Link State Routing (OLSR)[5]protocol is modified to support multiple ra-dios.OLSR propagates link information in the network.Since multiple paths have to be used simultaneously and path selection happens on a packet by packet basis,102030405060051015CAM (ms)A c h i e v e d T h r o u g h p u t (M b p s )(a)β=0.0(b)β=0.5(c)β=1.0Figure 4.Achieved throughput versus CAM for 100randomly selected pairs of paths.source routing is used to forward the packets.The mea-surement of ETT is done using the ETX and packet pair technique described in [7].The nodes in the network are homogeneous in that they support the same num-ber of radios.Each node periodically exchanges ETT information with each of its neighbors.The periodic ex-change of ETT information is infrequent,about every 20minutes,because there is no mobility in the network.Neighbor link information is propagated throughout the network every 30minutes.Multipath discovery is through a simple exponential exhaustive ing a breadth first search,we build all possible routes from a source to a destination and limit the maximum path length to six.λand γare com-puted 3for all possible multipath combinations between the two nodes;the multipath with the lowest CAM is se-lected.Even though the exhaustive search is inefficient,it serves our purpose to evaluate the metric.Because the neighbor link information is only updated in the network once every 30minutes,we limit the re-computation of paths to once every 15minutes.The load ratio of each set of selected paths is rounded to the closest integer value and packets are split on the paths in this ratio.The simulations consist of 100nodes equipped with IEEE 802.11a radios in a 2000m X 2000m area.Unless otherwise mentioned,we use three 802.11a radios per node tuned to orthogonal channels 34,42and 46.The channel assignment is thus a redundant scheme as in [1,7].For computing WCETT,we assign equal weight to channel diversity and path length.Hence,we select ηin Equation 6to be 0.5.The traffic consists of a single UDP stream split over the multiple paths in the chosen load distribution ra-tio.The maximum achievable throughput is obtained through multiple simulations,where each simulation in-crementally increases the bandwidth of the stream.−3−2−10123024681012Difference in Load on the PathsA c h i e v e d T h r o u g h p u t (M b p s )Figure 5.Impact of load variation on throughput.paths is varied from (1:4)to (4:1).We found that for more than 50%of the paths,the achieved throughput varied by at least 25%over the entire load distribution range.Figure 5shows the throughput variation over a chosen set of six paths.The x-axis represents the dif-ference in the load over the paths.For example,a load distribution ratio of (1:2)is shown as -1on the the graph.The load variation affects some paths more than the others.The predicted ratio as discussed inSection 4was found to lie within 10%of the peak value for about 78%of the multipaths.The results show that traffic distribution plays a role in determining the achieved throughput on a multipath combination.Hence it is crucial to distribute the flows in the right proportion.Figure 6.Throughput gain with CAM.5.3.Throughput GainWe next perform a throughput comparison of multi-path routing with single path routing.Single paths areselected using the WCETT metric.Figure 6illustrates the achieved throughput comparison between multipath routing and single path routing.The results represent the average throughput obtained between 100randomly se-lected node pairs in the network.The x-axis represents the path length of the single path selected with WCETT.We refer to CAM with β=0.5as “CAM”.When β=1.0,only the channel diversity between paths (λ)is taken into consideration during the selection of paths and when β=0.0,only the end-to-end characteristics of paths (γ)is taken into account.Thus we refer to the former case as the “Maximum channel diversity”met-ric and the latter as the “Minimum WCETT average”metric.The results show that the use of multiple paths provides a significant throughput gain primarily with ex-tremely short path lengths.Figure 7.Channel utilization versusthroughput.To investigate further,we evaluate the correlation be-tween the gain in throughput and the reduction in the per packet bottleneck channel consumption time with the use of multiple paths.λdenotes the bottleneck on mul-tipath.Let λs denote the single path bottleneck channel consumption time.If m denotes the number of channels on the single path,λs can be given by:λs =max 1≤i ≤mX i(11)We define the channel gain as follows:Channel Gain =λs −λPath Length (hops)A c h i e v e d T h r o u g h p u t (M b p s )Figure 8.Throughput gain with CAM.Beyond 0.25,the throughput gain with multipaths is sig-nificant.One explanation for this behavior is that when the gain in channel bandwidth due to the additional path is low,the extra contention introduced compensates for the gain in throughput and can reduce the end-to-end throughput.Hence it is better to use single path rout-ing when the gain in channel bandwidth is low.We now plot the throughput gain comparison with the restriction that multiple paths are used only when the channel gain is more than 0.25.Figure 8shows the results only for the subset of the paths for which multi-path routing is used with CAM.In this scenario,CAM (β=0.5)significantly outperforms both single path routing and the other βvalues.The average gain is about 50%.5.4.Impact of Orthogonal ChannelsWe next evaluate the impact of the number of orthog-onal channels on the achieved throughput.We use a re-dundant channel assignment scheme with one interface per channel.The number of channels is varied by chang-ing the number of radio interfaces per node.The aver-age throughput gain with two-path routing over single path routing is shown in Figure 9.Beyond six channels,more than two paths are required to exploit the band-width in the network.We observe that β=0.8provides higher gain than β=0.2when the number of channels is more than three.When the available channel diver-sity is low,the end-to-end characteristics of the paths (γ)play a greater role in limiting the achievable throughput.With more channels available,the channel diversity (λ)between the paths has to be maximized to achieve larger gains in throughput.Typically,nodes are equipped with only two or three interfaces.A more optimal channel assignment scheme [20,21]needs to be in place to exploit all the avail-able orthogonal channels with the radio.The multipath gain with such a channel assignment scheme depends onFigure 9.Throughput versus number of orthogonal channels.the connectivity in the network and the number of paths available between the nodes.We plan to investigate the gain of multiple paths with such channel assignments in the future.6.Related WorkMultipath routing has been an active research area in ad hoc networks for years.Techniques have been pro-posed to achieve load balancing,failure recovery,error resilience and increased end-to-end throughput in multi-hop wireless networks [8,13,15,16].These techniques use multiple paths to alleviate the issues that arise out of dynamicity in ad hoc networks.The issues could be link failures on the primary path,packet loss due to con-gestion or interference,path breaks and route changes due to node mobility and failures.All of these tech-niques achieve the benefits of error resilience and load balancing through spatial diversity provided by multiple paths.Video encoding [3,14]and multipath partition-ing techniques have been proposed to improve the end-to-end video quality in multihop networks.A number of multipath routing protocol solutions [9,11,12,17,22]have been proposed for ad hoc networks that discover and route packets on multiple paths between the source and destination.Our work assumes the presence of a static multi-hop mesh network with little dynamicity in the network.Node failures and link breaks are infrequent.We focus on increasing the end-to-end throughput with efficient utilization of the available frequency channels in the neighborhood of the nodes.The notion of ET T as the link metric captures packet errors and bandwidth on a link.Unlike earlier work,the emphasis is on better utilization of the channel diversity rather than spatial diversity.。
