Presentation is loading. Please wait.

Presentation is loading. Please wait.

1 SenMetrics’05, San Diego, 07/21/2005 SOSBRA: A MAC-Layer Retransmission Algorithm Designed for the Physical-Layer Characteristics of Clustered Sensor.

Similar presentations


Presentation on theme: "1 SenMetrics’05, San Diego, 07/21/2005 SOSBRA: A MAC-Layer Retransmission Algorithm Designed for the Physical-Layer Characteristics of Clustered Sensor."— Presentation transcript:

1 1 SenMetrics’05, San Diego, 07/21/2005 SOSBRA: A MAC-Layer Retransmission Algorithm Designed for the Physical-Layer Characteristics of Clustered Sensor Networks Qingjiang Tian and Edward J. Coyle Center for Wireless Systems and Applications (CWSA) School of Electrical and Computer Engineering Purdue University {tianq,coyle}@ecn.purdue.edu

2 2 SenMetrics’05, San Diego, 07/21/2005Outline  Background  SOSBRA Approach for Clustered Sensor Networks  Numerical Results  Optimal Contention Window  Conclusions

3 3 SenMetrics’05, San Diego, 07/21/2005 Introduction  Design for Energy Efficiency Through All Layers of the Protocol Stack  Cross-Layer Design to Improve Performance Need to avoid fragility  My Work: Physical-MAC Layer Interface Small Propagation delay in sensor net applications Opportunity to redesign Retransmission algorithms Physical Energy Efficiency MAC Network Application

4 4 SenMetrics’05, San Diego, 07/21/2005 Background  General – 802.11 MAC Layer CSMA/CA Collision Avoidance Binary Exponential Backoff Homogeneous peer-to-peer Designed for hidden nodes (RTS-CTS Handshake)  V. Bharghavan, “MACAW: A Media Access Protocol for Wireless LANS” All nodes can hear each other  Y. Kwon,etc, “A Novel MAC Protocol with Fast Collision Resolution for Wireless LANs” multiplicative-increase, linear-decrease  C. Wang,etc, “A new collision resolution mechanism to enhance the performance of IEEE 802.11 DCF,” contention window size is halved after c consecutive successful transmission

5 5 SenMetrics’05, San Diego, 07/21/2005 Motivation for My Work  IEEE 802.11 Distributed Coordination Function (DCF) Called WiFi Homogeneous, peer-to-peer Communications Binary exponential backoff & cross-stage collisions

6 6 SenMetrics’05, San Diego, 07/21/2005 Motivation for My Work  Clustering in Sensor Networks Clusterhead: central control, broadcasting, synchronization of other nodes Energy efficiency is a goal Increase throughput on the channel »Minimize collisions and idle time Very Small Propagation Delay fd SENSOR fd SENSOR fd SENSOR fd SENSOR fd SENSOR fd SENSOR fd SENSOR 100m

7 7 SenMetrics’05, San Diego, 07/21/2005 SOSBRA: Synchronized, One-Stage-Backoff Retransmission Algorithm  Assumptions One-hop cluster considered Traffic model: collect one packet from each node within the cluster We ignore the small propagation delay between sensor nodes and CH All nodes within one cluster can be synchronized to within 1 microsecond Synchronization beam – similar to ZigBee – starts “rounds” or retransmissions on the channel Nodes can sense each other’s activity

8 8 SenMetrics’05, San Diego, 07/21/2005 SOSBRA Approach 1.Each node that needs to either transmit or retransmit at the beginning of a round will chose a slot at random in a contention window of size W for its retransmission. 2.Nodes that transmit without collision are done. 3.Nodes in collisions in the current round will reschedule transmissions in the next round of W slots. 4.W is the same for every round.

9 9 SenMetrics’05, San Diego, 07/21/2005 SOSBRA vs 802.11 DCF A A AC Standard 802.11 DCF 1 2………………… W 1 2 ……………. W New Round A B SOSBRA-based 802.11 DCF Window 1 Window 2 B C B ABC

10 10 SenMetrics’05, San Diego, 07/21/2005 PerformanceAnalysis Performance Analysis N: Total non-CH nodes within the cluster W: fixed one stage contention window :Total time required to collect one packet from each node : The duration of a RTS collision : The duration of a data packet transmission

11 11 SenMetrics’05, San Diego, 07/21/2005 PerformanceAnalysis Performance Analysis  No collisions (1) (2)

12 12 SenMetrics’05, San Diego, 07/21/2005 PerformanceAnalysis Performance Analysis  N1 nodes succeed in the first round and all of remaining N2 nodes succeed in the second round,C1 collisions in the first round (3) (4)

13 13 SenMetrics’05, San Diego, 07/21/2005 PerformanceAnalysis Performance Analysis (5)(5) (6) General Case

14 14 SenMetrics’05, San Diego, 07/21/2005 Numerical And Simulation Results Fig.1 Numerical results for the probability mass function of, the total time to empty the cluster, for the SOSBRA-based 802.11 protocol. Here, N =50 nodes and W =120.

15 15 SenMetrics’05, San Diego, 07/21/2005 Numerical And Simulation Results Fig. 2 Simulations for the SOSBRA-based 802.11 protocol that show during empty the cluster for different contention window sizes. is the number of nodes in the cluster.

16 16 SenMetrics’05, San Diego, 07/21/2005 Numerical And Simulation Results Fig.3 Simulations for the SOSBRA-based 802.11 protocol that show the average channel throughput during the emptying the cluster for different contention window sizes. N is the number of nodes in the cluster.

17 17 SenMetrics’05, San Diego, 07/21/2005 Numerical And Simulation Results Fig. 4 Simulations determining the optimal contention window size for different for the SOSBRA-based 802.11 protocol

18 18 SenMetrics’05, San Diego, 07/21/2005 Numerical And Simulation Results Fig. 5 Simulations determining the minimum, for different cluster sizes for the SOSBRA-based 802.11 protocol.

19 19 SenMetrics’05, San Diego, 07/21/2005 Numerical And Simulation Results Fig. 7. : Simulations comparing the wasted-time before the cluster is emptied for the SOSBRA-based 802.11, Standard 802.11 DCF, and ZigBee with and without GTS.

20 20 SenMetrics’05, San Diego, 07/21/2005 Numerical And Simulation Results Fig. 8. : Simulations comparing total energy consumption to empty the cluster for the SOSBRA-based 802.11, Standard 802.11 DCF, and ZigBee with/without GTS. The energy consumption ratios used was idle:receive:send=1:2:2.5 11

21 21 SenMetrics’05, San Diego, 07/21/2005 Numerical And Simulation Results Fig. 9. Comparison between SOSBRA and TDMA-based approaches. Here, and a slot time is 10 microsecond in SOSBRA.

22 22 SenMetrics’05, San Diego, 07/21/2005  Probabilistic Approach Cost Function Cost results from two sources »The first is from the total idle slot W »The other one comes from possible collisions (7) Optimal Contention Window Size

23 23 SenMetrics’05, San Diego, 07/21/2005 Optimal Contention Window Fig.10. Numerical Results showing Cost Function Vs 1/W

24 24 SenMetrics’05, San Diego, 07/21/2005 Optimal Contention Window Fig.11. Comparison between simulation and analytical results

25 25 SenMetrics’05, San Diego, 07/21/2005 Optimal Contention Window Fig. 12. Average Total time obtained with from both simulation and analysis.

26 26 SenMetrics’05, San Diego, 07/21/2005 Large Number of Nodes if for very large N, We may approximate the total cost to be (11)

27 27 SenMetrics’05, San Diego, 07/21/2005 Large Number of Nodes Define (12) (13) (14)

28 28 SenMetrics’05, San Diego, 07/21/2005Conclusions  SOSBRA provides better performance in term of both time and energy compare to 802.11 DCF  Help minimize the multi-access interference (collisions) in design of physical access scheme, especially for CDMA approach  Our future work includes analysis of cross layer designs for wireless sensors with directional transmission capability physical layer improvements, including adaptive modulation schemes synchronization across a sensor network CDMA based optimization of PHY/MAC design

29 29 SenMetrics’05, San Diego, 07/21/2005 Derivations of Formulas  N: Total non-CH nodes within the cluster  W: fixed one stage contention window  :Total time required to collect one packet from each node  : The duration of a RTS collision  : The duration of a data packet transmission

30 30 SenMetrics’05, San Diego, 07/21/2005 Derivations of Formulas

31 31 SenMetrics’05, San Diego, 07/21/2005 Derivations of Formulas (7) (8) (9)

32 32 SenMetrics’05, San Diego, 07/21/2005 Derivations of Formulas Given N nodes and length W contention window, for each of the W Slots: 1.No nodes choose this slot………………………. 1.Only one node chooses this slot……………...... 1.More than one nodes choose this slot………… Cost Function: Cost results from two sources: total # of empty slots and possible collisions Minimize the Cost Function: (10)

33 33 SenMetrics’05, San Diego, 07/21/2005 Derivations of Formulas if for very large N, We may approximate the total cost to be (11)

34 34 SenMetrics’05, San Diego, 07/21/2005 Derivations of Formulas Define (12) (13) (14)


Download ppt "1 SenMetrics’05, San Diego, 07/21/2005 SOSBRA: A MAC-Layer Retransmission Algorithm Designed for the Physical-Layer Characteristics of Clustered Sensor."

Similar presentations


Ads by Google