センサネットワークにおける 協調型データ蓄積方式について Cooperative Buffering for Wireless Sensor Networks

1. センサネットワークにおける協調型データ蓄積方式についてCooperative BufferingforWireless Sensor Networks 高田 悠　萬代 雅希　渡辺 尚 静岡大学大学院情報学研究科 KDDI研究所にて発表 2008/12/19（金）

2. 1 Background (1/2) • Wireless sensor networks (WSNs) • Many tiny sensor nodes • Each node sends its data to sink • Resource restriction of a sensor node • Processor, Transreceiver, Buffer size, etc. • Battery-powered →Saving energy of each sensor nodes • Advanced monitoring using images or sound • Cheap CMOS sensor, Small Camera • Microphone node sink ・camera（lens radius is about 1mm） ・several million pixel KDDI研究所にて発表

3. 1 Background (2/2) • The examples of advanced WSNs • surveillance video camera system • habitat monitoring • The characteristics of data on our research • Conventional model • air temperature, degree of humidity, vibration → small • Our model • large size • burst data generation • no immediacy • Ex) DTN: Delay Tolerant Networks • Large data transmission • multi-hop transmission may not suitable for large data Goal: Large data transmission with low power consumption in Wireless Sensor Networks KDDI研究所にて発表

4. Cooperative Buffering with Mobile Sinks • using neighbor node’s buffer • multiple nodes buffer the data cooperatively until the sink’s arrival→node’s power saving, large data buffering 2 Wireless sensor networks with large amount of delay-tolerant data • each node buffer its data • the size of a node’s buffer is limited → overflow • Related Works（wireless sensor networks with mobile sinks） • Data Mules [R. Shah, S. Roy, S. Jain, W. Brunette., IEEE Workshop on Sensor Network Protocols and Applications (SNPA), 2003] KDDI研究所にて発表

5. 3.1 CB (Cooperative Buffering) • buffer the data with 1-hop neighbor nodes • the source node still have data・・・ • buffer the data with 2-hop neighbor nodes • sink collects their data node generates large amount of data （define as a source node） sink data each node’s data transmission to the sink is 1hop KDDI研究所にて発表

6. 3.2 Details of CB (1/2) <buffer division> • Saving region for data relay • Data relay regardless of the amount of buffering data <neighbor table> • node ID (node_ID) • hop count from a source node（HC: Hop Count） • remaining buffer level (RBL) • Total RBL of leaf node（L_RBL: Leaf RBL） If RBL = 0, then data relay start RBL (Remaining Buffer Level) buffering relaying buffered data node’s buffer KDDI研究所にて発表

7. 3.2 Details of CB (2/2) <selecting method of cooperative node> • ns→n3: DATA 1 ～ 50 • n3→ns: L_RBL = 20 • ns→n2: DATA 51 ～ 90 • n2 → ns : L_RBL = 0 • ns →n1: DATA 91 ～ 120 • n1 → ns : L_RBL = 60 • ns → n1 →n6: DATA 121 ～ 160 • ns → n1 →n5: DATA 161 ～ 180 • ns → n3 →n4: DATA 181 ～ 200 RBL: remaining buffer level L_RBL: total RBL of leaf node 7 8 6 40 20 5 1 40 30 2 s 50 3 4 20 data buffer region = 100 KDDI研究所にて発表

8. Problem S2 sends the data to distant node, needs more power 3.3 Multi-source CB (1/2) • Multiple neighbor source nodes • S2 starts data relay after S1 have finished CB The RBL of s1 , s2,n1 ,n2 ,n3 =0 n5 n2 n6 s2 s1 n3 source node n1 cooperatively buffering node n4 vacant node KDDI研究所にて発表

9. multiple buffer region 　→ acceptance of multi requests s2 3.3 Multi-source CB (2/2) • Solution • Dividing the buffer region into m buffer The division number of buffer regionm = 2 buffersource 2 buffersource 1 s1 n5 n2 transfer s2 s1 n3 The number of cooperative node increases as m increases n1 n4 settings of m is very important KDDI研究所にて発表

10. The relational expression of analyze the relation of C, D and 4.1 Evaluation 1 • Total power consumption： • Source node cooperates within h-hop neighbors • transmission power consumption • source node → cooperative nodes • cooperative nodes → mobile sink • receive power consumption of each nodes KDDI研究所にて発表

11. 4.2 Result 1 （the number of neighbor nodes: N = 5,transmission power consumption: Pt=12, receive power consumption: Pr=1.8） ※mote based C=1 h = 5 h = 4 h = 3 C=2 h = 2 h = 1 • total power consumption decreases as C increases • If C is big, the number of cooperative nodes decreases • According to D gets large, more power is required • source node needs to buffer data with far outer nodes C=5 D KDDI研究所にて発表

12. 5.1 Evaluation 2 (1/2) • Computer Simulation • Performance metric：total power consumption • Simulation Parameters • MAC protocol：CSMA/CA 2way • Transmission rate： zigbee • Power consumption： mote • Data generation： CBR • Generation interval： 1(s) KDDI研究所にて発表

13. 5.1 Evaluation 2 (2/2) • Simulation Environment • # of nodes: 36 • topology: random node • source node runs CBaccording to the amount of data • mobile sink arrives at the source nodeafter it has finished CB • cooperative nodes and the source nodesend data to a mobile sink directly • node’s buffer size： 512kB※mote • buffer region: 128 and 256 of 512kB mobile sink KDDI研究所にて発表

14. 128kB 256kB 5.2 Result 2 • CB’s power consumption lower than Multi-hop • CB can decrease message relays • The bigger buffer region leads to low power consumption • node has bigger buffer region can buffer more data KDDI研究所にて発表

15. 6 Summary and Future Work • Summary • CB (Cooperative Buffering) with mobile sink • Performance evaluation • CB outperforms conventional Multi-hop data transmission in terms of the total power consumption • Bigger buffer region leads to lower power consumption • Future Work • CB in Sparse WSNs • Setting of the number of division of buffer region (m) KDDI研究所にて発表

16. Thank you very much for your attention!!Yu Takada KDDI研究所にて発表

17. 移動シンクとなりうる移動体 • 移動における制約が比較的ゆるい移動体 • 人 • バイク, 自転車 • 動物（牧牛, 野鳥など） • 移動における制約が厳しい移動体 • バス, 電車, タクシーなどの自動車 • 船 • 飛行機 • 衛星 KDDI研究所にて発表