Node Reclamation and Replacement for Long-lived Sensor Networks Bin Tong, Wensheng Zhang, and Chuang Wang Department of Computer Science, Iowa State University Guiling Wang Department of Computer Science, New Jersey Institute of Technology IEEE SECON 2009

Outline Introduction Goal System Model Node reclamation and replacement The ARTS Scheme Performance Evaluation Conclusion

Introduction Wireless Sensor Network Power by batteries

Introduction Wireless Sensor Network Environmental sources Deployment Reclamation and Replacement

Introduction Motivation and Goal Providing a guaranteed Quality of Service A sufficient number of sensor nodes being alive Minimizing the overhead caused by the reclamation and replacement The travel distance of the MR is minimized in the long run

System Model System Architecture The system consists of a mobile repairman (MR), an energy station (ES) A WSN composed of groups of sensors surrounding the posts The MR traverses the network periodically to reclaim sensors having low or no energy and replace them with fully-charged sensors group

Assumptions Time is divided into phases of a constant length. A certain number of phases compose a round, the length of which is denoted as l. The MR visits each post at most once every round. A sensor has two modes: active and sleep. At the beginning of each phase, all sensors in each group should wake up and participate in the duty-cycle scheduling. Node reclamation and replacement Surveillance Number = 6 group … phase1phase2phase i Wait time

The ARTS Scheme Adaptive rendezvous-based two-tier scheduling scheme Local-Tier Scheduling Global-Tier Scheduling round j round j+1 1. Visit time 2. Number of sensors to be reclaimed/replaced residual energy Collects information for the round j+2 phase 1phase 2

The ARTS Scheme Local-Tier Scheduling Quality of Service is guaranteed ( ≧ N max ) Remaining energy in sensors to be reclaimed/replaced should be as small as possible Initial phase j Scheduling 1. N d 2. Surveillance number 3. t Scheduling phase j N d : the number of sensors in the group t: the number of remaining phases before the next replacement/reclamation

N max × tδ The ARTS Scheme Local-Tier Scheduling Quality of Service is guaranteed ( ≧ N max ) Remaining energy in sensors to be reclaimed/replaced should be as small as possible N d : the number of sensors in the group t : the number of remaining phases before the next replacement/reclamation e i : the amount of sensor i remaining energy δ: be the amount of energy consumedby an active sensor per phase Sorted L 1 = e i ≧ tδ L 2 = e i ＜ tδ Scheduling check remaining energy in L 2 m × tδ

The ARTS Scheme Local-Tier Scheduling Quality of Service is guaranteed ( ≧ N max ) Remaining energy in sensors to be reclaimed/replaced should be as small as possible L1L1 L2L2 L1L1 L2L2 L2L2 L1L1 δ=1, N max = 5 greedy scheduling policy: Fail 6 ≧ (5-3)3*1=6 2 ＜ (5-3)3*1=6

The ARTS Scheme Local-Tier Scheduling Controlled-greedy Algorithm δ=1, N max = 5 2 ＜ (5-3)3*1=6 x

The ARTS Scheme Global-Tier Scheduling The total travel distance of the MR is minimized The remaining energy of sensors to be replaced is minimized group i round j information of round j+1 visiting time and replacement number in round j +2 for group i

The ARTS Scheme Global-Tier Scheduling The total travel distance of the MR is minimized The remaining energy of sensors to be replaced is minimized Calculation of Replacement Numbers : Choose least energy sensors l : the length of a round τ : the lifetime of sensor if being active all the time (l/τ) × N max = 3 × N max Case: l >τ N max Case: l ≦ τ To guard against the worst case scenario when N max nodes are needed to be active all the time. jj+1j+2 time j=1,2: j >2:

The ARTS Scheme Global-Tier Scheduling The total travel distance of the MR is minimized The remaining energy of sensors to be replaced is minimized Calculation of the Travel Schedule for the MR Data Structure in round j+2 Table R Deadline e(i, p): the total residual energy in the sensors to be reclaimed/replaced in group g i at phase p

Data Structure in round j+2 G(V, E, W (V ), W (E), R)  V = {g i | 0 ≤ i ≤ n}  W (V) ={N r (1, j+2), · · ·, N r (n, j+2)} = (t 1, t 2, · · ·, t n ) ︰ a visiting time vector D = ︰ is the total traveling distance for the MR = (2, 1, 3, 4 ) g1g1 g2g2 g3g3 g4g4 N r (1, j+2) N r (2, j+2) N r (3, j+2) N r (4, j+2) g1g1 g2g2 g3g3 g4g4 N r (1, j+2) N r (2, j+2) N r (3, j+2) N r (4, j+2)

The ARTS Scheme Global-Tier Scheduling The total travel distance of the MR is minimized The remaining energy of sensors to be replaced is minimized Calculation of the Travel Schedule for the MR How decide the vector to minimize the weight g1g1 g2g2 g3g3 g4g4 N r (1, j+2) N r (2, j+2) N r (3, j+2) N r (4, j+2)

Global-Tier Scheduling The total travel distance of the MR is minimized The remaining energy of sensors to be replaced is minimized Calculation of the Travel Schedule for the MR g1g1 g3g3 g7g7 g6g6 g2g2 g4g4 ES Super Tour 1 Super Tour 2 M [10] [10] T. Liebling, D. Naddef, and L. A. Wolsey, “On the capacitated vehicle routing problem,” Mathematical Programming, 2003 g5g5

Performance Evaluation Simulation Parameter lifetime τ4000 minutes phase length10 minutes round l4000 minutes N max 8 sensors/group 1000 m 36 groups 16 sensors 40 sensors

Performance Evaluation Simulation Result Tradeoff between Residual Energy in Sensors to Be Reclaimed/Replaced and MR’s Travel Distance Comparison with Optimal Solution

Performance Evaluation Simulation Result Impact of MR’s Capacity Impact of Round Length

Conclusion A node replacement and reclamation (NRR) strategy and an adaptive rendezvous-based two-tier scheduling (ARTS) scheme to meet the challenges of designing an efficient WSN for long-term tasks Extensive simulations have been conducted to verify the effectiveness and efficiency of the ARTS scheme. Future Work Plan to explore other design choices for the NRR System, extend the current design of the ARTS scheme, and set up real testbed to evaluate the design.

Thank you~