March 20 th Sensor Web Architecture and Protocols

Transcription

1 March 20 th 2017 Sensor Web Architecture and Protocols Soukaina Filali Boubrahimi

2 Why a energy conservation in WSN is needed? Growing need for sustainable sensor networks Slow progress on battery capacity Target Habit Traffic Tracking Monitoring i Control sustainable sensor networks Border Control Infrastructure health Monitoring Space Monitor 2

3 BasicApproaches to Energy Conservation in WSNs Duty Cycling Data Driven Mobility Approach Topology Control Power Management Energy Efficient Data Acquisition Data Reduction Mobile Sink Synchronious Asynchronious In Network Processing Mobile Relay Receiver Initiated Data Compression Sender Initated Data Prediction Adaptive 3

4 A MAC X MAC LPL MAC 4

5 5

6 - Each node periodically (ts) wakes up and senses the channel. - If the channel is busy: stay awake for tw. - Upon receiving a packet, extend awake time for a certain period. - Send preambles or short strobes until the receiver wakes up. - ELSE node.sleep() Parameters: ts: the sleep interval. tw: awake time td: dynamically extended awake time τ : the overhead to sense the channel. 6

7 There are two cases for sending a packet in duty cycling protocols: 1. Preambled transmission: If the receiver is sleeping, the sender should wait until the receiver wakes up. 2. Non preambtransmission: The packet is transmitted led without preambles. 7

8 There are three sources of energy consumptions: 1. Radio on on time for sensing 2. Radio on time for receiving packets, including the time for receiving packets and channel sensing. 3. Radio on time for sending packets =>The proposed protocol estimates both energies to optimize to overall energy consumption 8

9 1. Designed a framework to qualitatively analyze the significant impact of traffic patterns and protocol dynamics in duty cycling protocols. 2. Proposed a light weight distributed duty cycling protocol design (LAD) which can achieve optimal energy efficiency with different data rates and protocol dynamics in real networks. 3. Implemented the protocol in TinyOS with 40 TelosB nodes. And a trace driven simulation of 1200 nodes in a real CitySee network deployed in a urban area 9

10 10

11 11

12 Radio on time for sending packets Energy Estimation: Estimate the number of non preamble packets is: Estimate the number of preamble packets is: 12

13 Average Energy Estimation: Energy consumption for channel sensing is denoted by τ Energy consumption for receiving packets: The energy consumption for radio on time for receiving ii packets is αe(l), where α is a coefficient for energy consumption. Energy consumption for sending packets: For each preambled transmission, the time ts/2. For each non preambled transmission, the time is negligible. Therefore, the expected energy consumption for sending packets can be calculate as βe(mp)ts/2 with β as a coefficient. 13

14 . (a) λ = (pkt/ms) (b) λ = 0.02 (pkt/ms) (c) λ = 0.1 (pkt/ms) For low data rates: Energy consumption increases when tw and td increases (consider having smaller td) For medium data rates: Energy consumption decreases when tw and td increases (consider having higher tw and td) For high data rates: Energy consumption decreases when td increases (consider having higher tw and td) 14

15 The design consists of 3 major components: Nt Network keti Estimation Parameter Optimization Adaptive Duty Cycling Protocol 15

16 16

17 s1 tw 0 td 0 ts tw 200 td 100 ts 17

18 幻灯片 17 s1 soukaina, 3/17/2017

19 Adaptive Duty Cycling Protocol Component: Each node adjust the ts, tw and td according to the parameter optimization component Each node send its schedule to its neighbors with maximum preamble length from neighbors ts The information is broadcasted to the network using piggybacking 18

20 The LAD protocol was compared with : TinyOS LPL MAC [1] (LPL) with default settings: (ts = 500ms, tw = 10ms and td = 100). TinyOS LPL MAC with minimal td value, td = 0 (LPLnoextending). Parameter optimization with X MAC [2]. A MAC [3], i.e., the most recent receiver initiated duty cycling protocol. Performance was compared from the following aspects: Duty cycle ratio, the percentage of radio on time. Average energy consumption per packet. Packet loss ratio. Adaption to different data rates. Detailed radio operations. 19

21 High Data Rate Low Data Rate 20 En ergy Consumption Duty Cycle ratio

22 High Data Rate Low Data Rate 21 Reliability of Packet Transmission

23 NodePosition Sleep IntervalDistribution => The protocol adapts to the data rate 22

24 Radio Operation for LPL MAC Radio Operation for LAD eceiver Re der Send 23

25 CitySee Network consists of 1200 nodes, each node send 4 data packets to the sink every 10 mins. TinyOS LPL is used with ts=512, td=10 and tw=10. Data Distribution Duty cycle y Improvement CDF 24

26 State of the art protocols cannot efficiently adapt to traffic and protocol dynamics. Thus they are not accurate and adequate to optimize the energy consumption In this paper, a practical adaptive duty cycling protocol that reduces the energy consumption is presented. The protocol minimizes the energy consumption per packet undervarious traffic rates andprotocol dynamics. The approach was evaluated on 40 TelosB nodes and a 1200 node network, the results show that the LAD approach can improve the performance by 28.2% 40.1%. 25

27 Poor usability of pre calculated sleep times in a network which h experiences a high h degree of network kdynamics. Little effort was dedicated to find an optimized packet length; while this parameter takes a long convergence time to determine. dt 26

28 [1] TinyOS LPL MAC. [Online]. Available: 2.x/doc/html/tep105.html [2] M. Buettner, G. V. Yee, E. Anderson, and R. Han, X mac: a short preamble mac protocol for duty cycled ldwireless sensor networks, in ACM SenSys, [3] Y. Sun, O. Gurewitz, and D. B. Johnson, Ri mac: a receiverinitiated asynchronous duty cycle mac protocol for dynamic traffic loads in wireless sensor networks, in ACM Sensys, [4] Wang, Jiliang, et al. "Sleep in the dins: Insomnia therapy for duty cycled sensor networks. " INFOCOM, 2014 Proceedings IEEE. IEEE,