1 / 32

Understanding Packet Delivery Performance in Dense Wireless Sensor Networks

Understanding Packet Delivery Performance in Dense Wireless Sensor Networks. Jerry Zhao & Ramesh Govindan SenSys ‘03. Motivation. WSNs can be deployed in harsh environment Measure packet delivery performance Spatio-temporal charasteristics of packet loss Environmental dependence

eyal
Download Presentation

Understanding Packet Delivery Performance in Dense Wireless Sensor Networks

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Understanding Packet Delivery Performance in Dense Wireless Sensor Networks Jerry Zhao & Ramesh Govindan SenSys ‘03

  2. Motivation • WSNs can be deployed in harsh environment • Measure packet delivery performance • Spatio-temporal charasteristics of packet loss • Environmental dependence • Medium scale (up to 60 Mica motes) indoor, habitat with moderate foliage, and open parking lot -> Implications for the design & evaluation of routing & MAC protocols

  3. Why packet delivery performance is important? • Determines energy efficiency & network lifetime • Poor packet delivery may degrade application performance & consume a lot of energy • Important for evaluating communication protocols • Experimentally verify WSN design principles, for example, low-power RF transceivers for multiple short hops • More energy efficient than a single hop over a long range • Spatial multiplexing

  4. Backgrounds: Some Wireless Communication Vagaries • Hidden node problem: Node A transmits to B, Node C cannot hear it and transmits to B -> Collision at B C A B

  5. Backgrounds: Some Wireless Communication Vagaries • Exposed node problem: Node B is transmitting to A, Node C has a packet intended for node D -> C cannot transmit, although it’s OK C D A B

  6. Backgrounds: Some Wireless Communication Vagaries • Multipath problem • A radio signal is reflceted by obstacles • Parts of the signal may take different paths to the sink, confusing the receiver Source: Wireless Lan, Multipath and Diversity http://www.cisco.com/en/US/tech/tk722/tk809/technologies_tech_note09186a008019f646.shtml

  7. Backgrounds: Some Wireless Communication Vagaries • Signal attenuation • Attenuation = (10/L) log10 (Pi/Po) where L is the distance, e.g., meter or km • dB/m or dB/km • Signal strength drops exponentially • Signal strength is proportional to 1/ra where r is the distance and 2 ≤ a ≤ 5

  8. Packet delivery performance • Physical layer • If there’s no interfering transmission, delivery perf is largely determined by a function of environment, physical layer coding scheme, individual receiver charasteristics (not a major factor) • MAC layer • Interfering transmissions contribute to poor perf. • Evaluate the efficacy of carrier sense and link layer retransmission

  9. Contributions • Experiments & observations • No new protocols or algorithms • Lack of the related work on delivery performance measurement in a medium scale WSNs (when the paper was published) • Although the results do not necessarily mean radio communications in WSNs are always like this, they provide important insights

  10. Key Results • Heavy-tailed distributions of packet losses • For example, in an indoor setting, half of the links experience more than 10% packet loss, and a third suffer more than 30% loss • Physical layer: Gray area within the communication range • Receivers suffer choppy packet reception • In some case, gray area is 1/3 of the comm. range • MAC layer: Packet loss is heavy-tailed • 50% - 80% comm. energy is wasted to overcome packet collisions & environmental effects • About 10% of links exhibit asymmetric packet loss

  11. Authors suggest • Topology control, via actual measurement of actual perf, needs to carefully discard poorly performing links or neighbors to whom asymmetric links exist • Packet level mechanisms, e.g., RTS/CTS, are not enough • Make decisions at the granularity of links to neighbors

  12. I. Packet delivery at the physical layer • Disable TinyOS MAC to measure pure packet delivery at physical layer • Vary three factors • Environments • Physical layer coding schemes • Transmit power settings

  13. Environment 1 • I:Indoor office building • 2m * 40m hallway • 60 motes placed in a line • 0.5m apart • 025m apart near the edge of the comm range • Removed some node from near the transmitter • Harsh due to significant multipaht reflection effects

  14. Environment 2 • H: 150m * 150m segment of a state park • Downhill slope with foliage and rocks • Multi-path problems due to foliage & rocks

  15. Environment 3 • O: 150m * 150m open parking lot • No obstacles • Multipath only due to ground reflections • Not much to sense

  16. Physical layer encoding scheme • SECDED (Single Error Correction and Double Error Detection) • TinyOS default • Convert each byte into 24 bits • Can detect 2 bit errors & correct one bit error • Manchester encoding • Convert a byte into 16 bits • Detect an error out of 2 bits • 4-bit/6-bit scheme (4bsb) • Encode one byte into 12 bits • Detect 1 bit error out of 6 bits

  17. Discrete control of transmit power in a mote • Three settings are considered • High (potentiometer 0) • Medium (potentiometer 50) • Low (potentiometer 90) • Potentiometer is an electric device with user-adjustable resistance

  18. Aggregate packet delivery performance • Pacekt loss with 4b6b coding, high Tx power -> Worst case pkt delivery perf. I H O

  19. Aggregate packet delivery performance • Packet loss vs Tx power in I, 4b6b coding • Observelower power improves dilivery perf considerably possibly due to the reduced comm range and multi-path problems H M L

  20. Aggregate packet delivery performance • Pkt loss vs coding schemes in I, high Tx Power • SECDED is much better for the cost of consuming more bandwidth than 4BSB and Manchester • Not much difference btwn 4BSB and Manchester

  21. Spatial Characteristics of Packet Delivery • How does reception rate vary with distance from the transmitter? • Gray area due to multipath problems Spatial profile of packet delivery: 4B6B, High Tx Power I H O

  22. Why servere multipath problem? • No frequency diversity • Motes use a single, narrow frequency band • How about emerging UWB (Ultra Wide Band) technology? • 3.1 – 10.6 GHz • Bandwidth > 500MHz • Data rate > 54Mbps • Low power

  23. Lessons • Selecting a shortest path simply based on the geographic distance or hop count is not sufficient! • Nodes need to carefully select neighbors based on the measured packet delivery perf!

  24. Signal strength & packet delivery • Try to answer a question: “Can signal strength by itself estimate link quality?” • Unfortunately, the answer is “NO” High Tx Power, I

  25. Coding Schemes • “Can sophisticated physical layer coding schemes mask the gray area?” • Not necessarily, SECDED has the lowest effective bandwidth -> Topology control to avoid pathological links in the gray area together with bandwith efficient coding scheme

  26. Spatial Correlation • “Are two receivers in their linear topology likely to see similar loss patterns?” • Significantly different correlation characteristics for different environments: I & O show noticeably higher correlated packet loss than H • At the physical layer, independent losses are a reasonable assumption I O H

  27. Temporal characteristics of packet delivery • Large variations in average reception rate and big standard deviations imply time varying packet losses

  28. II. Packet Delivery at the Medium Access Layer • TinyOS • CSMA/CA: Random back off upon carrier sense • Link layer ACK: Send 4 byte ACK to the sender • Authors added retransmission scheme • When there’s no ACK, retransmit up to 3 times

  29. Packet loss distribution under the Retransmission Scheme • Too many packet loss • 50% - 80% communication energy is wasted on repairing lost transmissions • Better MAC, e.g., S-MAC, B-MAC, Z-MAC, is required

  30. Asymmetry in packet delivery • Asymmetry in wireless communication is well known, but the extent is not • Topology control should control pathological links

  31. Conclusions • Performed experiments to understand packet delivery perf in dense sensor network deployments • Quantify the prevalence of gray area • Mostly “observations” • “Causes” for phenomena are not for sure • Most of them are conjectures, guesses, etc. • Still an open issue 

  32. Questions?

More Related