An Energy-Efficient and Reliable Multi-hop Cluster-based Wireless Sensor Networks

1. An Energy-Efficient and Reliable Multi-hop Cluster-based Wireless Sensor Networks • 指導教授：王國禎 博士　　學生：徐逸懷 • 國立交通大學資訊工程研究所 • 行動計算與寬頻網路實驗室

2. Outline • Introduction • Related work • Proposed energy-efficient and reliable multi-hop cluster-based (ERMC) WSNs • Performance evaluation • Evaluation metrics • Simulation results • Conclusion • References

3. Introduction (1/3) • Wireless Sensor Networks (WSNs) are comprised of large • amount of energy-constrained sensor nodes, which • communicate with each other and have low processing ability • Since it is impossible to recharge the battery of sensor nodes, • how to reduce energy consumption of sensor nodes to extend • the network life time of WSNs becomes a critical issue [1]

4. Introduction (2/3) • In hierarchical clustering protocols, each cluster consists of a • cluster head (CH) and cluster members (CM). CH is used to • be a leader of a cluster which is responsible for data • aggregation/fusion, sleep scheduling, etc [2] • The nodes having higher density factor which means having • more neighbors are elected as cluster head so that the energy • consumption can be minimized as well as network life time • can be maximized [3]

5. Introduction (3/3) • After the construction of a cluster, if one node is not involved • in any cluster, which is called isolated node [1]. The isolated • node may cost much more energy consumption than others • because it needs to find the path to relay data for the sink on • its own • Once the WSN structure is well-established, the inter-cluster • communication between CHs is necessary. In other words, • CHs use multi-hop communication with each other through its • CM inside the cluster

6. Related work • Comparison of proposed ERMC with other protocols

7. Proposed ERMC protocol (1/16) • Network architecture [3] • The sink is fixed and located far away from the sensor nodes • The sink knows node location, node identity (ID), and initial energy of each node • The sensor nodes are energy constrained with a uniform initial energy allocation • The nodes have fixed transmitted power • Each node senses the data at a fix rate and always has data to send to the sink • All nodes are capable of moving

8. Proposed ERMCprotocol (2/16) • The operation of ERMC is divided into rounds • Each round consists of four phases, which are cluster • formation phase, isolated node assisted phase, steady phase, • and Cross-CHs route construction phase

9. Proposed EMC-MAC protocol (3/16) • Cluster formation phase • the sink selects cluster head candidates from a set of alive nodes whose energy level is higher than the average energy level • Sink selects CHs from the candidate set depending on which have higher sensor density that is the number of closest neighbors • For calculating the density of each CH candidate, we use the following equation • Where X is any alive node, 1…P is a set of candidate nodes, and • is the absolute distance between X and i

10. Proposed EMC-MAC protocol (4/16) • Network Topology

11. Proposed EMC-MAC protocol (5/16) • Node density of each candidate node

12. Proposed EMC-MAC protocol (6/16) • After the calculation of the node density of each candidate node, we use the following cost function to select CH • where N is the number of neighbor of the candidate node j • S is a set of candidate nodes, is the square distance between • node i and node j • FD refers to the density factor which is the node density of a • candidate node

13. Proposed EMC-MAC protocol (7/16) • We calculate for CH set selection • Where

14. Proposed EMC-MAC protocol (8/16) • Isolated node assisted phase

15. Proposed EMC-MAC protocol (9/16) • Steady phase • The mobile agent needs to move forward and back between its CH and isolated node for data collecting

16. Proposed EMC-MAC protocol (10/16) • The CH periodically broadcasts a request-reply message for its mobile agent members to checks if all the mobile agents are back in the cluster • When all mobile agents are back, CH broadcasts a data-collection message for all its mobile agents to request them for collecting data from their neighbors • After mobile agent collects data from its neighbors and performs data aggregation, the mobile agent transmits its data to the CH • CH schedules all the members which has not transmitted its data and broadcasts a TDMA scheduling message for all these members

17. Proposed EMC-MAC protocol (11/16) • Cross-CHs route construction phase • If there is any isolated node in a direction to the sink, CH selects the isolated node to be a gateway • If there is more than one isolated node in the direction to the sink, the CH chooses the isolated node whose corresponding mobile agent has highest energy level as a gateway • If there is no isolated in the direction to the sink, the CH needs to select a proper gateway to relay data to the sink • CH multicast gateway-selection message to those cluster member it the direction to the sink • These members independently starts a backoff timer to send a CatewayClaim message to its CH

18. Proposed EMC-MAC protocol (12/16) • We define the normalized hop cost from CH1 to CH2 through node A • where v is the immediate neighbor of node A which belong to • is ClusterDegreeof a cluster member A as the number of • different clusters its immediate neighbor nodes belong to • the residual energy level of node v • Let denotes the average normalized hop cost from CH1 to CH2 that can be accessed through node A

19. Proposed EMC-MAC protocol (13/16) • Let denotes the backoff delay for a cluster member a • where is a time slot unit • returns a random value between 0 and which is used to • reduce the potential simultaneous transmission

20. Proposed EMC-MAC protocol (14/16) • Gateway selection

21. Proposed EMC-MAC protocol (15/16) • CHs start to transmit their data to the sink by the well-established path • They will piggyback the energy level of nodes within their cluster into the transmitted data to the sink • The sink will check whether the is less than • If true, the sink recalculates the optimal set of CHs and • broadcasts the beacon message toannounce the starting of • next round

22. Proposed EMC-MAC protocol (16/16) • Flowchart of node behavior in each round

23. Performance evaluation - metrics (1/4) • Simulation parameter [1,3,4]

24. Performance evaluation - metrics (2/4) • Simulation parameter [1,3,4]

25. Performance evaluation - metrics(3/4) • Energy consumption model [5]

26. Performance evaluation - metrics(4/4) • Control message transmission • the total number of control message transmissions to establish the routes to the sink • Broadcasting transmission • the total number of transmissions when a source CH broadcasts a single packet to the sink

27. Performance evaluation - simulation results (1/6) • Control message transmission with respect to Network Size

28. Performance evaluation - simulation results (2/6) • Control message transmission with respect to Cluster Size

29. Performance evaluation - simulation results (3/6) • Broadcasting transmission with respect to Network Size

30. Performance evaluation - simulation results (4/6) • Broadcasting transmission with respect to Cluster Size

31. Performance evaluation - simulation results (5/6) • Number of nodes alive

32. Performance evaluation - simulation results (6/6) • Energy dissipation per round

33. Conclusion • Conclusion • Proposed an energy-efficient and reliable multi-hop cluster- • based (ERMC) WSNs • Using node density and minimum distance between CHs • with their neighbors as the metrics to choose CHs, • exploiting mobile agent to solve isolated node problem, and • utilizing the node mobile agent saved as a gateway or using • an energy-efficient mechanism to select gateway to relay • data to the sink effectively, ERMC has the following benefits: • Its control message transmission is 61.33% lower than CBCDACP • Its broadcasting transmission is 68.18% lower than • CBCDACP

34. References • [1] Yun-Sheng Yen; Ruay-Shiung Chang; Sin-Lung Ke; "An Energy-Efficient • Clustering Protocol for Wireless Sensor Networks," Computer and Network • Technology (ICCNT), 2010 Second International Conference on , pp.18-22, 23- • 25 April 2010. • [2] A. A. Abbasi and M. Younis, “A survey on clustering algorithms forwireless • sensor networks,” Computer Communications, vol. 30, no. 14-15, pp. 2826– • 2841, 2007. • [3] JannatulFerdous, M.; Ferdous, J.; Dey, T.; "Central Base-Station Controlled Density Aware Clustering Protocol for wireless sensor networks," Computers and Information Technology, 2009. ICCIT '09. 12th International Conference on, pp.37-43, 21-23 Dec. 2009. • [4] Long Cheng; Das, S.K.; Di Francesco, M.; Canfeng Chen; Jian Ma; "Scalable and Energy-Efficient Broadcasting in Multi-Hop Cluster-Based Wireless Sensor Networks," Communications (ICC), 2011 IEEE International Conference on , pp.1-5, 5-9 June 2011. • [5] Heinzelman, W.B.; Chandrakasan, A.P.; Balakrishnan, H.; , "An application-specific protocol architecture for wireless microsensor networks," Wireless Communications, IEEE Transactions on , vol.1, no.4, pp. 660- 670, Oct 2002.