Disruption Tolerant Shell (SYS 13)
暂无分享,去创建一个
Center for Embedded Networked Sensing Disruption Tolerant Shell Martin Lukac, Lewis Girod, Deborah Estrin CENS System Lab – http://research.cens.ucla.edu Introduction: Data Collection and System Management in Challenged Networks Meso American Subduction Experiment • Extensive: 500 Km from Acapulco through Mexico City to Tampico • Dense: 1 sensor every 5-10 Km • High bandwidth: data acquisition rate: 3 x 24 bit channels at 100Hz each • Online and reliable: semi real-time (on the order of days), reliable data delivery to UCLA for analysis • Online system management: query state, change configuration, update binaries • Application driven topology: application determines sensor placement. Infrastructure does not Software Requirements • Data delivery – Bandwidth driven –Bandwidth: 20-40 of MB per day per station –Latency: get the data eventually, but reliably –Many to one routing • System management – Latency driven –Bandwidth: usually less than 10’s of KB’s –Latency: as fast as possible –One to all routing and back 50 standalone Caltech sites 62 wirelessly connected UCLA sites Problem Description: End - to - End Tools Fail at Critical Times • Frequent unpredictable disconnections –Rainy season: sites flood (some 24x7) and trees grow –Wind/weather: misaligned antennas –Equipment malfunction: amps burn, voltage regulators break • Poor and unstable links –Connectivity is a secondary concern for site selection –Stretched links highly susceptible to weather and environment • Human effort is a critical resource –Installation, maintenance, protection • Data delivery and system management techniques Distribution of Effective Bandwidth Measurements for Typical Links designed for wired or always-on-wireless do not work well –Typical tools use TCP to create and maintain an end to end session to deliver a stream of data over multiple hops –These tools expect reliable links with low latencies • Patterns of poor links, disconnections, and disruptions –Difficult to obtain and maintain end-to-end connections –Intermittent end-to-end connections insufficient for KB/s required bandwidth and latency J-F (5 similar) E-B (5 similar) M-I (3 similar) D-C (0 similar) Bandwidth Variability 13 node section Proposed Solution: Disruption Tolerant Shell Data Delivery: DTN • Use Delay Tolerant Networking techniques • Buffer data into hour long bundles (1-3 MB) • Deliberate one hop bundle transfer • Path to sink determined by best ETX • Improvement over end-to-end –Not affected by path disconnections –Keeps retrying on single link instead of full path –Continual ‘progress’ being made towards sink –More efficient use of bandwidth in face of disconnections and bottlenecks Fraction DTS Results - Cuernavaca • Compared latency of DTS to parallel ssh • DTS is faster 90% of the time, comparable the rest of the time • DTS reaches 100% of nodes –ssh requires retries from the source node • Latency can vary by day, but DTS always faster or comparable to ssh Latency of DTS vs. end-to-end ssh for node G Fraction Successful ssh trials (94.2%) Successful DTS trials (100%) DTS Network Service: StateSync • StateSync: Reliable and efficient Seconds publish-subscribe mechanism A PUBLISH System Management: DTS • Implements a broadcast dissemination protocol • Existing management tool: remote shell (ssh) –Published data is hop-scoped Commands Responses • Modified management tool: Disruption –DTS publishes commands and responses one hop Tolerant Shell (DTS) • Works well for applications that require: –Asynchronous remote shell to all nodes in SYNCHRONIZE –Reliable delivery network simultaneously –Have a few Kbytes of data to share –Provides node management capabilities when –Data lifetime is long compared to system latency requirements end-to-end connections are unavailable or fail B PUBLISH –Suitable for DTN since it does not use end-to-end connections –Ensures that commands will succeed: as long • StateSync data model: tables of key value pairs as there is eventually a connection between a Commands Responses node and any other node that already has the –DTS has a command table and response table command • Logging mechanism DTS features • Guaranteed in order execution from source node –Do not republish whole table: only send changes to tables SYNCHRONIZE • Safe recovery from reboots and crashes –More efficient use of bandwidth in face of disconnections • Implicit feed back on nodes and links: spot • Retransmission protocol bottlenecks, dead nodes C –Keeps retrying on individual links PUBLISH • Execute a command on individual nodes –Not affected by path disconnections • Push a file to all nodes Commands Responses –No overhead of end-to-end connection –Distribute new script or component UCLA – UCR – Caltech – USC – CSU – JPL – UC Merced