Resilient Approach for Energy Management on Hot Spots in WSNs Fernando Henrique Gielow Michele Nogueira Aldri Luiz dos Santos {fhgielow,michele,aldri}@inf.ufpr.br NR2 – Federal University of Paraná IFIP/IEEE IM2011 Dublin, May 23th, 2011 Outline Introduction and motivation Related work CEA – Cluster-based Energy Architecture RRUCR Definition of scopes Clustering Initial backbone creation Cluster-heads rotations Data gathering & Routes maintenance Evaluation Conclusion Introduction and motivation Sensor nodes: constrained resources Low processing and storage capabilities Limited lifetime Applications Surveillance/monitoring systems Data gathering applications Introduction and motivation Traffic patterns n to n n to 1 Unequal traffic distribution Data gathering applications Areas burdened with higher traffic rates Introduction and motivation Hot Spots There are more important nodes in the network Unequal traffic distribution Areas burdened with higher traffic rates Mitigation through sink mobility, biased deployment, unequal clustering Related work Hot Spot mitigation Sink mobility Biased deployment The nodes close to the sink will change eventually Unpractical in the majority of scenarios E.g. [Thanigaivelu and Murugan, 2009] Manually deploying more nodes near the sink Unpractical in the majority of scenarios E.g. [Wu and Chen, 2006] Unequal clustering Smaller clusters near the sink More routes to reach the sink E.g. [Chen et al, 2009] CEA: Cluster-based Energy Architecture Generic behavior Cluster-based Hot spot mitigation Intra/inter clusters energy management Route management The RRUCR protocol Rotation Reactive Unequal Cluster based Routing Cluster-based multi-hop protocol for networks with n to 1 traffic pattern Data gathering applications Unequal clusters to mitigate hot spots Dynamic maintenance of routes Operations of RRUCR Definition of scopes Clustering Initial backbone creation Cluster-heads rotations Data gathering & Routes maintenance RRUCR Definition of scopes Transmission powers ordered and indexed The sink covers those powers 4 4 3 4 2 3 4 3 2 2 1 Potence indexused used Potence indexPotence used index 2 to cover by the sink by the 3sink toby cover current node to nearest node Potence most distant node reach the sink index limit to this operation RRUCR Clustering Unequal sized Hot spot mitigation (funneling of routes) Previously defined scope power Balanced quantity Avoids dense areas Still cover all nodes RRUCR Initial backbone creation Process initialized by the sink Process carried by cluster-heads Update route and forward message once RRUCR Cluster-heads rotations Balance internal energy consumption Prolong network lifetime Generate broken links when selecting farther nodes Force route update of the node that rotated ( ) and of the nodes which used the previous CH ( ) RRUCR Data gathering & Routes maintenance CHs route the data to the sink in a multi-hop way That message carries a field which indicates the distance from the node up to the sink Used to update routes If obligatory and a node with shorter distance found Reactive approach, with low overhead Evaluation Parameters 1000x1000m 700 nodes Initial energy between 0.9 and 1.1 J 1% prob. of generating 32 bytes of data at each 0.1s 5000s Radio parameters set according to Mica2 3 scenarios Without failures With failures close to the sink With failures far from the sink ns-2.30 simulator 35 simulations – interval of confidence of 95% Compared to UCR 700 650 600 550 500 450 400 350 300 250 200 RRUCR UCR 4500 Death of the first node (s) Total energy Evaluation 0 1000 2000 3000 4000 RRUCR UCR 4000 17% 3500 13% 3000 2500 2000 1500 5000 Close Far None Failures Time (s) 24 70 RRUCR UCR 60 Number of clusters Number of rotations 21% 50 40 30 20 10 RRUCR UCR 21 18 15 12 9 6 3 0 0 0 1000 2000 3000 Time (s) 4000 5000 1 2 3 4 5 Hops to the sink 6 7 Evaluation RRUCR 30 25 20 Number of dead nodes Number of dead nodes UCR DS <= 40 40 < DS <= 80 80 < DS <= 160 160 < DS 15 10 5 0 3000 4000 Time (s) 5000 30 25 20 DS <= 40 40 < DS <= 80 80 < DS <= 160 160 < DS 15 10 5 0 3000 4000 Time (s) 5000 Evaluation Data delivery rate (%) RRUCR RRUCR UCR 60 50 0.9 0.8 None Close Far 0.7 0 5 10 15 20 25 20 25 Rotations 40 30 UCR 20 10 0 0 1000 2000 3000 Time (s) 4000 5000 Data delivery rate (%) Number of rotations 70 1 1 0.9 0.8 None Close Far 0.7 0 5 10 15 Rotations Conclusion Address the Hot spot impacts on WSN RRUCR Less deaths close to the sink Improve network lifetime Balanced quantity of clusters Increased network lifetime in 21.36%, when compared to UCR Increased data delivery rates Work extended to a generic architecture (CEA) Future work Operations that check the integrity on WSN links More complex route maintenance A TinyOS implementation of RRUCR Project page with more information www.nr2.ufpr.br/~fernando/rrucr/ Source code available under LGPL ns-2.30 simulation Installation script Doubts? Email for contact: fhgielow@inf.ufpr.br