Connected Point Coverage in Wireless Sensor Networks using Robust Spanning Trees IEEE ICDCSW, 2011 Pouya Ostovari Department of Computer and Information Siences Temple University Philadelphia, Pennsylvania, USA Mehdi Dehghan Department of Computer Engineering and Information Technology Amirkabir University of Technology Tehran, Iran Jie Wu Department of Computer and Information Siences Temple University Philadelphia, Pennsylvania, USA
Outline Introduction Related work Goal Assumption Proposed approach Simulation Results Conclusion
Introduction Coverage is one of the most important challenges in the area of sensor networks – Area coverage – Boundary coverage – Point coverage
Introduction The area coverage problem (Areas of Interest, AoI) – aims at covering the whole area. – Sensors are deployed to maximize the covered area.
The boundary coverage problem (Lines of Interest, LoI) – aims at detecting intrusion on a given area. – Sensors have to form a dense barrier in order to detect each event that crosses the barrier. Introduction USA Intruder
Introduction The point coverage problem (Points of Interest, PoI) – aims at monitoring specific points in the field of interest. MuseumCampusMilitary
Related work I. Cardei and M. Cardei, “Energy-efficient connected-coverage in wireless sensor networks,“ International Journal of Sensor Networks, Vol. 3, No. 3, pp , May 2008 Sink Targets Sensing nodes Waiting nodes
Related work Every sensors calculates the back-off time – Residual energy – The number of uncovered targets it can cover I. Cardei and M. Cardei, “Energy-efficient connected-coverage in wireless sensor networks,“ International Journal of Sensor Networks, Vol. 3, No. 3, pp , May 2008 Sink Targets Sensing nodes Waiting nodes
Related work Minimum Spanning Tree(MST) – Prim's Algorithm GraphMST
Related work Minimum Spanning Tree(MST) – Prim's Algorithm I. Cardei and M. Cardei, “Energy-efficient connected-coverage in wireless sensor networks,“ International Journal of Sensor Networks, Vol. 3, No. 3, pp , May 2008 Sink Targets Sensing nodes Waiting nodes Virtual links
Related work Every sensing nodes finds the path between virtual link(target,target’parent) I. Cardei and M. Cardei, “Energy-efficient connected-coverage in wireless sensor networks,“ International Journal of Sensor Networks, Vol. 3, No. 3, pp , May 2008 Sink Targets Sensing nodes Waiting nodes Virtual links Physical links
Related work Other nodes can go to sleep I. Cardei and M. Cardei, “Energy-efficient connected-coverage in wireless sensor networks,“ International Journal of Sensor Networks, Vol. 3, No. 3, pp , May 2008 Sink Targets Sensing nodes Waiting nodes
Related work The sink does not have many branches. – In the case of failure, we lose a significant amount of data. I. Cardei and M. Cardei, “Energy-efficient connected-coverage in wireless sensor networks,“ International Journal of Sensor Networks, Vol. 3, No. 3, pp , May 2008 Sink Targets Sensing nodes Waiting nodes
Cycle Formation Related work I. Cardei and M. Cardei, “Energy-efficient connected-coverage in wireless sensor networks,“ International Journal of Sensor Networks, Vol. 3, No. 3, pp , May 2008 Targets Sensing nodes Virtual links Physical links 3 2 1
Cycle Formation Related work I. Cardei and M. Cardei, “Energy-efficient connected-coverage in wireless sensor networks,“ International Journal of Sensor Networks, Vol. 3, No. 3, pp , May 2008 Targets Sensing nodes Virtual links Physical links 3 2 1
Goal To maintain connectivity, we propose a method to construct a balanced tree which is more robust against failure We consider and avoid cycle formation which might happen during converting virtual trees to physical trees.
Assumption Sensors and targets are stationary Each sensor knows the location of all targets and the sink
Our Approach Phase 1 : Sensing Nodes Selection Phase 2 : Relay Nodes Selection Avoiding Cycle Formation
Sensing Nodes Selection A set of efficient sensing nodes is selected to cover all the targets in the field – Residual energy – The number of uncovered targets it can cover
Sensing Nodes Selection Each sensor computes a waiting time – When the waiting time of sensor s u is finished, it is selected as a sensing node – The shorter the waiting time of sensor is, the more the priority among the others is
Sensing Nodes Selection The waiting time of a sensor s u is computed by the equation: E’ u : the residual energy of sensor s u E : the initial energy of sensor M : the number of targets in the network W 1 : the maximum waiting time α,β : weight values TargetS u : targets which are in sensing range of node s u and have not been covered by any sensing node yet T’ u : the waiting time which sensor s u has passed u v
Sensing Nodes Selection The waiting time of a sensor s u is computed by the equation: E’ u : the residual energy of sensor s u E : the initial energy of sensor M : the number of targets in the network W 1 : the maximum waiting time α,β : weight values TargetS u : targets which are in sensing range of node s u and have not been covered by any sensing node yet T’ u : the waiting time which sensor s u has passed u v
Sensing Nodes Selection When the waiting time of a sensor s u is finished u v r3r3 r2r2 r4r4 r1r1 r5r5 E’ u : the residual energy of sensor s u E 1 : active sensor consumes E 1 energy for sensing E 2 : active sensor consumes E 2 energy for communication TargetS u : targets which are in sensing range of node s u and have not been covered by any sensing node yet
Sensing Nodes Selection S u acts as the supervisor of all the targets in the set TargetS u u v r3r3 r2r2 r4r4 r1r1 r5r5
Sensing Nodes Selection Sink Targets Sensing nodes Waiting nodes
Sensing Nodes Selection Sink Targets Sensing nodes Waiting nodes
Relay Nodes Selection Step 1 : Make a virtual tree (MST) based on the set of targets and sink Step 2 : Compute the new cost for all of the target by using VRST(Virtual Robust Spanning Tree) algorithm Step 3 : Construct the new virtual tree(VRST) Step 4 : Convert the virtual tree to a physical tree of sensors Extend : MVRST(Modified Virtual Robust Spanning Tree) algorithm
Relay Nodes Selection Step 1 : Make a virtual tree (MST) based on the set of targets and sink Step 2 : Compute the new cost for all of the target by using VRST(Virtual Robust Spanning Tree) algorithm Step 3 : Construct the new virtual tree(VRST) Step 4 : Convert the virtual tree to a physical tree of sensors Extend : MVRST(Modified Virtual Robust Spanning Tree) algorithm
Relay Nodes Selection Prim's Algorithm – Cost = edge’s length Sink Targets Sensing nodes Waiting nodes
Relay Nodes Selection Sink Targets Sensing nodes Waiting nodes Virtual links Prim's Algorithm – Cost = edge’s length
Relay Nodes Selection Step 1 : Make a virtual tree (MST) based on the set of targets and sink Step 2 : Compute the new cost for all of the target by using VRST(Virtual Robust Spanning Tree) algorithm Step 3 : Construct the new virtual tree(VRST) Step 4 : Convert the virtual tree to a physical tree of sensors Extend : MVRST(Modified Virtual Robust Spanning Tree) algorithm
Relay Nodes Selection In VRST algorithm, cost is computed as follows : which λ is a function of the depth of a vertex : hop count : number of hops to the sink weight of path i : sum of the edge’s cost which connect vertex i to the sink ε 1 : depth of the MST h i : depth of vertex i z i,j : length of edge (i,j) hihi λiλi r1r1 r2r2 r3r3 r4r4 - (2) - (3) - (4) hihi λiλi i j
Relay Nodes Selection Step 1 : Make a virtual tree (MST) based on the set of targets and sink Step 2 : Compute the new cost for all of the target by using VRST(Virtual Robust Spanning Tree) algorithm Step 3 : Construct the new virtual tree(VRST) Step 4 : Convert the virtual tree to a physical tree of sensors Extend : MVRST(Modified Virtual Robust Spanning Tree) algorithm
Relay Nodes Selection MSTVRST
Relay Nodes Selection Step 1 : Make a virtual tree (MST) based on the set of targets and sink Step 2 : Compute the new cost for all of the target by using VRST(Virtual Robust Spanning Tree) algorithm Step 3 : Construct the new virtual tree(VRST) Step 4 : Convert the virtual tree to a physical tree of sensors Extend : MVRST(Modified Virtual Robust Spanning Tree) algorithm
Relay Nodes Selection Select some relay nodes to connect the supervisors of targets Targets Sensing nodes Waiting nodes Virtual links u v (t i, π(t i )) titi π(ti)π(ti) r
Relay Nodes Selection Every sensing node s u broadcasts a control message Targets Sensing nodes Waiting nodes Virtual links u v titi π(ti)π(ti) RELAY_REQ Location of sensing node s u Location of destination target π(t i ) The maximum distance from sensing node s u to the supervisor of target π(t i ) r
Relay Nodes Selection Targets Sensing nodes Waiting nodes Virtual links u v titi π(ti)π(ti) RELAY_REQ Location of sensing node s u Location of destination target π(t i ) The maximum distance from sensing node s u to the supervisor of target π(t i ) r If r is not the supervisor of π(t i ) and RELAY_REQ
Relay Nodes Selection Targets Sensing nodes Waiting nodes Virtual links u v titi π(ti)π(ti) r RELAY_REP 1. Relay sensor closest to the node s u 2. The first relay node which delivered the message 3. Relay node with the most residual energy RELAY_REP
Relay Nodes Selection Targets Sensing nodes Waiting nodes Virtual links u v titi π(ti)π(ti) r Physical links
Relay Nodes Selection Sink Targets Sensing nodes Waiting nodes Virtual links Physical links
Relay Nodes Selection Step 1 : Make a virtual tree (MST) based on the set of targets and sink Step 2 : Compute the new cost for all of the target by using VRST(Virtual Robust Spanning Tree) algorithm Step 3 : Construct the new virtual tree(VRST) Step 4 : Convert the virtual tree to a physical tree of sensors Extend : MVRST(Modified Virtual Robust Spanning Tree) algorithm
Relay Nodes Selection In VRST algorithm, cost is computed as follows : which λ is a function of the depth of a vertex : hop count : number of hops to the sink weight of path i : sum of the edge’s cost which connect vertex i to the sink ε 1 : depth of the MST h i : depth of vertex i z i,j : length of edge (i,j) - (2) - (3) - (4)
Relay Nodes Selection In MVRST algorithm, cost is computed as follows : which λ is a function of the depth of a vertex : hop count : number of hops to the sink weight of path i : length of edge(i,j) ε 1 : depth of the MST h i : depth of vertex i z i,j : length of edge (i,j) - (2) - (3) - (4)
Relay Nodes Selection VRSTMVRST
Avoiding Cycle Formation Targets Sensing nodes Virtual links Physical links 3 2 1
Avoiding Cycle Formation Targets Sensing nodes Virtual links Physical links 3 2 1
Simulation Results EnvironmentMATLAB Square field500m*500m Number of sensor500 Number of target50 Sensing energy consumption in a range of 50m 60mW/s Communication energy consumption in a range of 80m 60mW/s
Simulation Results Effect of sensing range on energy consumption : R c = 120m
Simulation Results Effect of sensing range on the maximum data loss : R c = 120mEffect of sensing range on the average data loss : R c = 120m
Simulation Results Effect of communication range on the maximum data loss : R s = 70mEffect of communication range on the average data loss : R s = 70m
Simulation Results Average depth : R s = 70m
Conclusion Propose a point coverage mechanism in addition to two connectivity mechanisms Our approach has less data latency and energy consumption
Sensing Nodes Selection Sink Targets Sensing nodes Waiting nodes Virtual links Physical links
Sensing Nodes Selection Sink Targets Sensing nodes Waiting nodes Virtual links Physical links
Sensing Nodes Selection-VRST Sink Targets Sensing nodes Waiting nodes Virtual links Physical links
Sensing Nodes Selection-VRST Sink Targets Sensing nodes Waiting nodes Virtual links Physical links
Sensing Nodes Selection-MST Sink Targets Sensing nodes Waiting nodes Virtual links Physical links
Sensing Nodes Selection-MST Sink Targets Sensing nodes Waiting nodes Virtual links Physical links
Sensing Nodes Selection-MVRST Sink Targets Sensing nodes Waiting nodes Virtual links Physical links
Sensing Nodes Selection-MVRST Sink Targets Sensing nodes Waiting nodes Virtual links Physical links