SNU INC Lab SNU INC Lab Introduction Sensor networks It is important to point that a node in the network has essentially two different tasks: Sensing its environment and processing the information. Forwarding traffic as an intermediate relay in the multi- hop path Significant energy savings are only obtainable by putting the node in sleep mode, essentially disconnecting it from the network and changing the topology
SNU INC Lab SNU INC Lab Introduction Topology Management The goal of topology management is to coordinated the sleep transitions of all nodes, while ensuring adequate network connectivity, such data can be forwarded efficiently to the data sink. Transfer state : Assume the network has data to forward. Monitoring state: No data needs to be forwarded to the data sink. Sensor networks often the vast majority of time in the monitoring state.
SNU INC Lab SNU INC Lab SPARSE TOPOLOGY MANAGEMENT Basic concept The sensor network is in the monitoring state the vast majority of its lifetime. STEM (Sparse Topology and Energy Management) Dual Frequency Wakeup plane : f1 Data plane : f2
SNU INC Lab SNU INC Lab THEORETICAL ANALYSIS Setup Latency(Ts) The statistics of Ts ( For T > T R x) P(Ts = B 1+2 ) = (T R x – B 1 ) / T P(Ts = kB 1+2 ) = B 1+2 / T (k = 2…K) P(Ts = (K+1)B 1+2 ) = (T - KB T R x + B 1 + B 1+2 )/ T The average setup latency per hop
SNU INC Lab SNU INC Lab THEORETICAL ANALYSIS Energy Savings. ignoring t data. the general relationship between setup latency and the relative energy gain
SNU INC Lab SNU INC Lab PERFORMANCE EVALUATION Simulation Setting R = 20m, L = 79.27m, N = 100 L beacon = L response = 144bits, B 1+2 = 150ms, T R x = 225ms Send 20 information packets of 1040bits to the data sink with an inter-packet spacing of 16 seconds t* = 320 seconds
SNU INC Lab SNU INC Lab PERFORMANCE EVALUATION Average Setup Latency The average setup latency per hop as a function of the wakeup period T. Eq(6) Average setup latency
SNU INC Lab SNU INC Lab PERFORMANCE EVALUATION Relative Energy Savings Without STEM With STEM Gain = 1 - Relative energy savings versus The total observation interval t
SNU INC Lab SNU INC Lab COMBINING STEM AND GAF GAF Behavior Average numbers of node in a grid is M M = M / Θ Θ is the fraction of used grid Interaction of STEM and GAF. Theoretical energy – setup delay tradeoff
SNU INC Lab SNU INC Lab CONCLUSION STEM ( topology management technique ) trades off power savings versus path setup latency in sensor networks. STEM integrates directly with other topology management schemes such GAF, and results in energy savings.