• Share
  • Email
  • Embed
  • Like
  • Save
  • Private Content


Flash Player 9 (or above) is needed to view presentations.
We have detected that you do not have it on your computer. To install it, go here.

Like this presentation? Why not share!

Spot Symposium 2009 Ksn Radio Stack



Invited Talk

Invited Talk
Sun SPOT Symposium 2009, London



Total Views
Views on SlideShare
Embed Views



1 Embed 2

http://www.linkedin.com 2



Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
Post Comment
Edit your comment

    Spot Symposium 2009   Ksn Radio Stack Spot Symposium 2009 Ksn Radio Stack Presentation Transcript

    • Markus Bestehorn Stephan Kessler Andreas Leppert pp Sven Meisinger g KSN Radio Stack: Reliable Wireless Communication for Data- intensive Applications in Sensor Networks Sun SPOT Symposium 2009 London www.kit.edu
    • Data-Intensive Application in WSN? Introduction Wireless Sensor Network (WSN) Data-Intensive Large number of battery p g y powered devices Sun St k S Stack energy is valuable! KSN Radio Stack Exchanging data using wireless communication Design Targets User can access network through base station Su SPOT Sun S O SHP Routing Countless Applications Other Layers Medical Care, Environment Monitoring, Object Tracking… Demo Comparison What is data-intensive application in WSN? KSN 2 variations: Team Nodes generate Data g Acitivities Data must be transported to user / base station Research Amount of data accumulates on the way Q&A Base station has to distribute data to the whole network 2 examples follow Markus Bestehorn Slide 2
    • Example 1: Data Gathering Introduction Data Gathering Applications Data-Intensive A large number of nodes g Sun St k S Stack Generating / Measuring data (continously) KSN Radio Stack Design Targets Sending measured data back to base station SHP Routing Other Layers Demo Comparison More Sensor Nodes KSN Team Base Acitivities Station Research Q&A Despite C D it Compression/Aggregation: i /A ti Messages of several KB have to transmitted reliably Large part of result is lost if 1 message is lost! Markus Bestehorn Slide 3
    • Example 2: Over-the-Air Deployment Introduction Testbeds Data-Intensive Evaluate your software y Sun St k S Stack Test and debug software KSN Radio Stack Design Targets SHP Problem: Testing-Cycle very cumbersome Routing Other Layers 1. Connect SPOT to USB Demo 2. Put S O s into pos o u SPOTs o position Comparison 3. Start testing / evaluation KSN 1. Errors in software? Team 2 Find error, fix code 2. error Acitivities Research 4. Collect SPOTs and start with step 1 Q&A Solution: Fixed infrastructure connected to nodes Not possible/feasible or all deployments Distribute software over-the-air (OTA) ( ) Markus Bestehorn Slide 4
    • Example 2: Over-the-air Deployment (cont‘d) Introduction Requires large files / messages to be transmitted Data-Intensive Q y Query Processor suite file ~500 KB Sun St k S Stack Libraries ~700 KB KSN Radio Stack Design Targets „Large“ Number of Hops SHP In I our testbed of 30 SPOT max. 9 Hops b d f SPOTs H Routing Other Layers Demo Any testbed requires transmission of large Comparison messages over multiple hops KSN Team Acitivities Research Q&A Markus Bestehorn Slide 5
    • Requirements of data-intensive Applications Introduction Communication software (stack) must Data-Intensive Deliver messages reliably g y Sun St k S Stack Cope with relatively large messages KSN Radio Stack Design Targets Multiple Hops between sender and receiver node SHP Particularly hard in P ti l l h d i sensor networks t k Routing Other Layers Expect nodes to fail during transmission Demo React to topology changes due to external influences, e.g., p gy g g Comparison Weather KSN Closing/Opening of doors Team Acitivities Research SPOTs are shipped with a communication stack Q&A Why not use this stack? Markus Bestehorn Slide 6
    • Testing the Sun Stack (1) Introduction Setup for evaluation Data-Intensive 2-5 nodes in a line, i.e., 1-4 hops , , p Sun St k S Stack 1 specified node sends a message of varying size KSN Radio Stack RadioStream was used Design Targets SHP After sending node computes a hash value of the message sending, Routing 1 specified node receives message Other Layers When receiving completes, receiver computes hash value over Demo received message Comparison Transmission successful, if both hash values are equal KSN Code can be obtained from our homepage: Team Acitivities htt // i d i k l h d /KSN http://www.ipd.uni-karlsruhe.de/KSN Research Q&A Sender Receiver Markus Bestehorn Slide 7
    • Testing the Sun Stack (2) Introduction Evaluation was conducted with the „Blue“ release (current) Data-Intensive Sun St k S Stack KSN Radio Stack Design Targets SHP Routing Other Layers Demo Comparison KSN Team Acitivities Research Q&A Markus Bestehorn Slide 8
    • Motivation for a new stack Introduction Sun stack is reliable and fast for small messages Data-Intensive Almost no errors for message sizes below 1KB g Sun St k S Stack Unreliable for large messages or high number of hops KSN Radio Stack Design Targets Main reason: Design Goals for the Sun Stack SHP Low P L Power Routing Other Layers Efficient usage of low bandwidth available Demo Design not aimed at „ g „data-intensive“ applications pp Comparison KSN Team Messages are transmitted reliably Acitivities over 1 h hop Research 802.15.4 layer works Q&A Errors must occur above Design new stack above MAC Markus Bestehorn Slide 9
    • Agenda Introduction Motivation: Why do we need another stack? Data-Intensive What does „data-intensive“ in sensor networks mean „ Sun St k S Stack Demo and reliability evaluation data KSN Radio Stack Design Targets KSN Radio Stack SHP Design D i targets & additional f ddi i l features Routing Other Layers The core of the KSN Stack Demo Optional KSN stack layers p y Comparison Possibilities for Optimization / Improvements KSN Reliability evaluation and Demo of the KSN stack Team Comparison: Sun Stack vs KSN Stack vs. Acitivities Research The KSN Team Q&A Projects Research Agenda Markus Bestehorn Slide 10
    • Design Targets - Reliability Introduction Transmission of large messages over multiple hops must Data-Intensive work reliably y Sun St k S Stack Problems encountered during transmission must be fixed inside KSN Radio Stack the communication stack if possible Design SHP Handling of errors should not be mixed with application code Routing Expect nodes to fail during transmission Other Layers External influences change communication behaviour of nodes Demo Weather Comparison Closing doors KSN Team Sender must know if transmission failed on intermediate Acitivities nodes End-to-End ACK Research Q&A ACK Sender Receiver Markus Bestehorn Slide 11
    • Design Targets - Extendability Introduction Different applications have different demands Data-Intensive Run Multiple Stacks p p parallely on one SPOT y Sun St k S Stack Adding / implementing new layers KSN Radio Stack Design Network Management might require another stack than Query SHP Processing Routing Stack with less complex interfaces for student experiments, Other Layers more sophisticated stack to gather the result of their Demo experiments Comparison KSN Team Acitivities Research Q&A Markus Bestehorn Slide 12
    • Further Design Targets Introduction Compatibility with 802.15.4 Data-Intensive Sun St k S Stack Interface Compatibility Interface-Compatibility to Sun Stack Users have written applications using the Sun Stack KSN Radio Stack Design Migrating to our stack should not require rewriting the SHP application Routing Link-Quality-awareness Other Layers Demo Probability for failures increases, as link quality decreases Comparison Communication hardware on SPOTs allows estimations about KSN link quality Team Example: Quality-Aware Routing Layer p y g y Acitivities Research Q&A Markus Bestehorn Slide 13
    • Optimization is not a Design Target Introduction Communication stacks for sensor nodes are usually Data-Intensive optimized regarding energy usage p g g gy g Sun St k S Stack Nodes are usually battery-powered KSN Radio Stack Design Reliability is the most important step towards efficiency SHP Users query a sensor network to gain information U t kt i i f ti Routing demand for information has to be satisfied (efficiently) Other Layers Demo If failures prevent delivery of information Comparison Energy used to acquire information until failure occured is wasted! Users might retry Failure likely to occur again more waste! KSN Team Optimization is the step after functionality is ensured p p y Acitivities Research “We should forget about small efficiencies, say about 97% of Q&A the time: premature optimization i th root of all evil.” th ti t ti i ti is the t f ll il ” (Donald Knuth) Markus Bestehorn Slide 14
    • Foundation for Reliable Communication Introduction Basic Operation of 802.15.4: Data-Intensive Sending a p g packet of 127 byte y Sun St k S Stack Receiver is a node within the radio range of the sender KSN Radio Stack Design Two variants of the Basic Operation: SHP Broadcast: R B d Receiver address i „blank“ i dd is bl k“ Routing No acknowledgement Other Layers Demo Unicast: Receiver address identifies node Comparison If transmission of 127 byte was successful, acknowledgement is sent KSN Team If there was a failure, no acknowledgement is received and a Acitivities i i timeout expires. Research Reliable communication is based on this basic operation Q&A First Step: Design communication protocol for messages larger than 127 byte Markus Bestehorn Slide 15
    • SHP – Single Hop Protocol Introduction Single Hop Protocol (SHP) Layer provides interface for Data-Intensive Unicast of messages of arbitrary size g y Sun St k S Stack Broadcast of messages of arbitrary size KSN Radio Stack M Design Preparation for sending (unicast or broadcast) SHP L t Message M h Let M have a size of b b t i f bytes Routing 1 2 3 4 Decompose M into f fragments Other Layers Demo Assign a sequence number s={1…p} g p Comparison to each fragment C KSN Compute checksum C from M Team Sending process for unicast and broadcast different Acitivities Research starting with broadcast Q&A Markus Bestehorn Slide 16
    • Single Hop Protocol - Broadcast Introduction Two types of packets Data-Intensive BroadcastData: Contains fragments g Sun St k S Stack of 104 Byte, used for first f-1 fragments KSN Radio Stack BroadcastEnd: Contains last fragment Design SHP with sequence number s=f s f Routing Possible failures: Other Layers Unexpected sequence number Demo Comparison Timeout while waiting for next/last fragment KSN Team Limitations (current implementation) ( p ) Acitivities Sequence Number is 16-bit long Research max. Broadcast ~ 6 MB Q&A sufficient for SPOTs Markus Bestehorn Slide 17
    • SHP – Possible Broadcast Improvement Introduction Problem: Broadcast still inherently unreliable Data-Intensive Sender broadcasts f fragments to surrounding nodes g g Sun St k S Stack Some receivers only receive k<f fragments KSN Radio Stack Design Receiver has spent energy, but message is incomplete SHP receiver discards all received fragments energy wasted! fragments, Routing Possible improvement (not yet implemented): Other Layers Demo Each receiver uses sequence numbers to determine, which k Comparison fragments are missing KSN If k is sufficiently small compared to f, ask sender to resend Team missing fragments with unicast Acitivities Open questions: Research Caching mechanisms for broadcast messages on sender Q&A What is „sufficiently small ? sufficiently small“? Markus Bestehorn Slide 18
    • Single Hop Protocol - Unicast Introduction SHP - Unicast Protocol Data-Intensive 1. Send UnicastRequest to receiver q Sun St k S Stack Contains Checksum C KSN Radio Stack Size b of Message M in bytes Design receiver can abort transfer SHP Routing 2. Send f fragments to receiver Other Layers Receiver has to acknowledge each Demo fragment g Comparison Unacknowledged fragments are KSN retransmitted after a timeout Team 3. Receiver checks if data is not Acitivities disrupted using checksum C Research Q&A Markus Bestehorn Slide 19
    • SHP – Unicast Improvements Introduction Evaluation has shown: Data-Intensive Data rates with Sun stack higher than with KSN stack g Sun St k S Stack Comparable only for small messages due to reliability KSN Radio Stack Design Reason for lower data rate: SHP Acknowledgment f each f A k l d for h fragment Routing congestion Other Layers Demo Less payload due to larger sequence numbers Comparison Ideas for improvements: „TCP/IP-like ACK“ KSN Aggregate ackowledgements and send 1 packet to Team acknowledge several received fragments Acitivities Increase number of packets between ackowledgements Research depending on link quality Q&A Markus Bestehorn Slide 20
    • Summary Single Hop Protocol Introduction SHP offers reliable delivery to 1-hop neighbors Data-Intensive Broadcast Sun St k S Stack Unicast KSN Radio Stack Design Message size SHP Bounded to 6 MB f b d B d d for broadcast SPOT has h Routing 512KB of Other Layers Theoretically unbounded for unicast VM defines boundaries for max. array length RAM Demo Comparison Layer still not fine-tuned KSN large room for improvements! Team Acitivities Research Design targets so far? Q&A Extendability not yet 802.15.4 compatiblity done Reliability only 1 hop Markus Bestehorn Slide 21
    • Extendability – Multiple Stacks, 1 SPOT Introduction All communication is based on reliable 1-hop communication Data-Intensive SHP usable as a foundation Sun St k S Stack Idea to allow multiple stacks KSN Radio Stack Design Each Stack is assigned a globally unique ID SHP we used a simple byte value Routing max. 256 different stacks Other Layers Special byte indicates stack ID Demo Comparison Each d E h node uses a Di Dispatcher t t h to Pass incoming data to correct stack depending on stack ID KSN Team Pass outgoing data to SHP Layer Acitivities SHP Dispatcher has 4 functions Research Register / unregister new Q&A stack Broadcast / unicast data using SHP Markus Bestehorn Slide 22
    • Extendability – More Stack-Layers Introduction All new layers must implement LayerInterface Data-Intensive Methods for broadcast / unicast of data from upper layers pp y Sun St k S Stack Method, which takes incoming data from lower layers or the KSN Radio Stack SHP Level Dispatcher Design SHP Abstract Layer implementation is provided with our software Routing New Layers can be implemented with very few lines of code Other Layers Demo Example of a AODV Routing Layer follows now Comparison KSN Team Acitivities Research Q&A Markus Bestehorn Slide 23
    • Example Layer – Multi-Hop Routing Introduction Up to this point, only 1-hop communication covered Data-Intensive Example for a new Layer: Routing Layer p y g y Sun St k S Stack Used Routing Mechanism: KSN Radio Stack Design Ad-Hoc On Demand Distance Vector Routing (AODV) [2] SHP Routing Other Layers Sun Stack also includes AODV implementation, but Demo No (working) End-to-End ACKs for Multi-Hop Unicasts ( g) p Comparison Only 1 entry per Receiver no backup routes KSN Both problems affect reliability Team Acitivities will be addressed in the Research following Q&A Markus Bestehorn Slide 24
    • AODV Routing Mechanism Introduction Basics of AODV: Data-Intensive Sender S wants to send message to Receiver R g Sun St k S Stack S broadcasts Route Request (RREQ) KSN Radio Stack Design Nodes receiving RREQs rebroadcast if SHP RREQ has not been broadcasted before Routing Node already has a route to R stored in routing table Other Layers Receiver R receives the RREQ Demo Route Report (RREP) is sent back to S Comparison Message is sent through shortest / fastest-available route KSN Team Acitivities RREQ Research Q&A RREQ RREQ RREQ RREP ACK ACK ACK Sender Receiver Markus Bestehorn Slide 25
    • AODV – Routing Table Format Introduction Routing Table without backup entries Data-Intensive 1 entry for each receiver R y Sun St k S Stack If transfer fails, new RREQ is broadcast by sender S KSN Radio Stack Design Problem: RREQ very small message (1 packet) SHP might be transmitted reliably, but larger messages fail reliably Routing same Route will be returned Other Layers transfer might fail again for larger message energy Demo wasted! Comparison Solution: KSN Several entries for 1 receiver R Team Acitivities Use depth-first search to transmit Research larger message to R Q&A Can be adapted to use quality information! Markus Bestehorn Slide 26
    • AODV – End-to-End ACK Introduction SHP sends ACK if message is transmitted correctly Data-Intensive Sun St k S Stack Problem: Intermediate node fails while sending KSN Radio Stack Design Intermediate node has sent ACK already, but message is lost! SHP Solution: Receiver sends End-to-End ACK to sender Routing Other Layers Sender waits for End-to-End ACK Demo Length o timeout de e e g of eou determined us g hop cou o routing table ed using op count of ou g ab e Comparison KSN Team SHP-ACK Acitivities Research Sender End-to-End ACK Receiver Q&A Design targets so far? g g Extendability done 802.15.4 compatiblity done Reliability done Markus Bestehorn Slide 27
    • Other Layers Introduction Reducing Data Size Data-Intensive Compression layer p y Sun St k S Stack Can be used with abitrary compression algorithm KSN Radio Stack We use a modified ZIP algorithm Design SHP Should be used above routing layer Routing Avoids decompression on every hop Other Layers Demo Protocol Dispatcher Comparison Applications can use different protocols KSN on top of single stack! Team Examples: p Acitivities PC-like with ports Research LowPan Q&A Markus Bestehorn Slide 28
    • KSN Radio Stack Demo Introduction Message Size: 50 KB Data-Intensive Tested up to 1 MB with more than 10 hops p p Sun St k S Stack 4 Hops KSN Radio Stack Design Output Power -30 SHP Difference to demo before Routing Other Layers „Simulation“ of node failures by removing/exchanging nodes Demo S u a ed „failure“ Simulated „ a u e will cause de ay delay Comparison Message delivered despite topology changes KSN „Application“ has not been changed Team Acitivities Research Q&A Sender Receiver Markus Bestehorn Slide 29
    • Comparison of both Stacks Introduction Sun Stack Data-Intensive Very fast and efficient for small messages y g Sun St k S Stack use for experiments where „every second counts“ KSN Radio Stack Unreliable if Design SHP Messages become large Routing Messages must be transmitted over a large number of hops Other Layers Other transmissions from other SPOTs run parallely Demo Was not shown here Comparison IPv6 compatible KSN KSN Radio Stack Team Acitivities Reliable for large message sizes and large number of hops Research Up to 1 MB Q&A And 15 hops tested without losing messages More overhead slower Use for experiments where „counting bytes“ is sufficient Currently not IPv6 compatible Markus Bestehorn Slide 30
    • Agenda Introduction Motivation: Why do we need another stack? Data-Intensive What does „data-intensive“ in sensor networks mean „ Sun St k S Stack Demo and reliability evaluation data KSN Radio Stack Design KSN Radio Stack SHP Design D i targets & additional f ddi i l features Routing Other Layers The core of the KSN Stack Demo Optional KSN stack layers p y Comparison Possibilities for Optimization / Improvements KSN Reliability evaluation and Demo of the KSN stack Team Comparison: Sun Stack vs KSN Stack vs. Acitivities Research The KSN Team Q&A Projects Research Agenda Markus Bestehorn Slide 31
    • Who is who @ KSN? Introduction Stephan Kessler Data-Intensive Student Assistant Sun St k S Stack KSN Serialization & Stack KSN Radio Stack Design SHP Andreas Leppert Routing Other Layers Student Assistant Demo KSN S ac , Que y Dissemination S Stack, Query sse a o Comparison KSN Team Sven Meisinger Acitivities Student Assistant Research KSN Serialization, Simulator Q&A Markus Bestehorn PhD student Project Head Markus Bestehorn Slide 32
    • KSN – Other Projects Introduction Management Application Data-Intensive Controlling large testbeds of SPOTs g g Sun St k S Stack Software deployment uses KSN Stack KSN Radio Stack Design Support for experiments, e.g., collect data SHP Serialization for SPOTs S i li ti f SPOT Routing Other Layers Serialization used in JDK to store/send objects Demo not supported by Squawk VM Comparison Pure Java serialization KSN Testbed Team 30+ nodes deployed at the IPD Acitivities Research Real network for experiments Q&A Many more… Homepage: www.ipd.uni-karlsruhe.de/KSN Markus Bestehorn Slide 33
    • KSN – Sensor Databases Introduction Research Topic: Sensor Databases Data-Intensive Problem: Every new application for WSN requires complete y pp q p Sun St k S Stack development process KSN Radio Stack Design & implement new hardware Design Design & implement new software SHP Routing very costly Other Layers Idea: Provide database-like interface for WSN Demo Comparison Advantage: U same software for new WSN application, only Ad t Use ft f li ti l change query KSN Team Example: SPOT 1 SPOT 2 Acitivities Temp: 25.3°C Temp: 23.4°C Research Hum.: 78% Hum.: 77% Light: 124 Light: 120 Q&A Measurement stored distributedly in tables Use SQL to query data, e.g., SELECT nodeID, temp FROM Sensors Size of result depends on query data-intensive! Markus Bestehorn Slide 34
    • KSN – Current Research Topics Introduction Approximate Query Processing Data-Intensive Reduce result quality to save energy q y gy Sun St k S Stack Different quality parameters possible KSN Radio Stack Precision of result Design SHP Security Routing Consistency Other Layers Query Distribution Demo Comparison Broadcast of query very energy consuming KSN Use structure of query to choose distribution strategy Team try to reach all relevant nodes, but not all nodes! Acitivities Spatio-Temporal query processing in WSN Research Moving object DBs provide operators to query object movement Q&A Only for historical data stored in a DB server Idea: Try to provide these operators for object tracking in WSN Markus Bestehorn Slide 35
    • Thank you for your attention! Introduction Data-Intensive Sun St k S Stack Questions? KSN Radio Stack Design SHP Routing Other Layers Demo References Comparison No. Reference KSN [1] MICA2 Sensor Node, http://www.xbow.com/ Team Acitivities Perkins, Royer: Ad h on-demand di t P ki R Ad-hoc d d distance vector routing; M bil C t ti Mobile Computing ti [2] Systems and Applications, 1999 Research Q&A Markus Bestehorn Slide 36