Power-Efficient Rendez- vous Schemes for Dense Wireless Sensor Networks En-Yi A. Lin, Jan M. Rabaey Berkeley Wireless Research Center University of California, Berkeley, USA Adam Woliz Telecommunication Networks Group (TKN) Technische Universitaet Berlin, Germany
Outline Introduction Rendezvous schemes between sensor nodes Synchronous Scheme Asynchronous Scheme Pseudo-Asynchronous Scheme MAC and Physical Layer Power Model Performance analysis of TICER & RICER Conclusions
Introduction Wireless sensor network characteristics Large number of nodes Nodes are energy limited Light traffic, short packets Power efficient can be achieved by putting nodes to sleep as much as possible Challenge Arrange simultaneous on-time for nodes wishing to communicate Rendezvous schemes
Introduction Rendezvous scheme The realm of the routing scheme Determining which node(specific/potential) a given node should rendezvous with in order to forward a packet Focus of this paper Focus exclusively on various rendezvous schemes and their power consumption characteristics, not on the most appropriate routing scheme for WSNs
Introduction Perspectives Analyze the cycled receiver scheme Based on a realistic model of the physical layer and the underlying hardware Take into account the effect of channel fading
Rendezvous schemes between sensor nodes Three categories of rendezvous schemes Synchronous Ye et al., “S-MAC”, 2002 Asynchronous Pseudo-Asynchronous Schurgers, “STEM”, 2002
Synchronous scheme A group of nodes in vicinity scheduled to periodically wake up at the same time
Synchronous Scheme Advantage Once setup, can use various handshake schemes Drawbacks Hard to accomplish Cause a major overhead
Asynchronous Scheme Nodes have the capability of waking up one another on demand Each node has two radios Normal radio : for data packet, mostly powered off Reactive radio : for control packet, always powered on with ultra low power ; can waken up by receiving a ‘wakeup signal’ from other nodes
Asynchronous Scheme Advantage The lowest power dissipation Drawbacks Require extra hardware Require some small standby power
Pseudo-Asynchronous Scheme Nodes establish rendezvous on demand, while using an underlying periodic wakeup scheme – (cycled receiver) Beaconing approach — the willingness to communication Have to satisfy certain performance requirements, like throughput and delay Transmitter or receiver initiates the rendezvous This paper To determine how the parameters influence the power consumptions To understand how to achieve a proper tradeoff between performance and power
Pseudo-Asynchronous Scheme TICER (Transmitter Initiated CyclEd Receiver) RICER (Receiver Initiated CyclEd Receiver)
Pseudo-Asynchronous Scheme Multiple potential receivers TICER The expected number of RTS’s transmitted from the source node until one of the potential receivers wakes up decreases RICER The expected waiting period of the source node for one of the potential receivers to wake up is shorter
Pseudo-Asynchronous Scheme Collisions in TICER and RICER A single control channel and a single data channel Not affect power consumption But does not completely eliminate collision For TICER data collisions may occur but with a small probability in WSN scenarios For RICER More node may respond with a data packet if receive a wakeup beacon Random delay
Pseudo-Asynchronous Scheme Advantage No scheduling or time synchronization required Simple Less overhead : power efficient Traffic distributed in time Fewer collisions Less overhearing Feasible with currently available hardware components Reactive radio is not commercially available yet
MAC and Physical Layer Power Model MAPLAP model : MAC and Physical LAyer Power model
MAPLAP Model A transceiver node is modeled as 5 states Transmit (TX) Receive (RX) Acquire (AQ) Monitor (MN) Idle (IL) AQ : the compute and power intensive timing phase and/or frequency acquisition phase MN : carrier sense
MAPLAP Model Parameters in each state Power consumption (P) Depends on PHY Probability (Δ) Depends on MAC/system parameters The average power consumption per node can be expressed as P total = Δ TX P TX +Δ RX P RX +Δ AQ P AQ + Δ MN P MN +Δ IL P IL
Performance analysis of TICER & RICER Parameters Control packet : 30 bits Data packet : 200 bits Data rate : 80 kbps Nodes average distance : 10 m Bit Error Rate (BER) : Channel fading T on = 60/date rate T l = 30/date rate Neighbors : 6 Same T for all node Propagation time negligible Realistic numbers from measurement P TX = 4.15mW P RX = 1.75mW P MN = 1.75mW P AQ = 2.00mW P IL = 0
Performance analysis of TICER & RICER Power consumption total TX RX AQ MN
Performance analysis of TICER & RICER Power consumption total TX RX AQ MN There exists an optimal T (wakeup period) assuring minimal power consumption Monitor power is very significant
Performance analysis of TICER & RICER Traffic load (with optimal T) TICER performs better at higher loads RICER : three way handshake TICER : four way handshake RICER has a lower chance of having an unsuccessful exchange of data RICER is more robust to fading
Performance analysis of TICER & RICER Traffic load (with non-optimal duty cycle) If we can not adapt optimal T, it is better to optimize T to a heavier traffic load
Performance analysis of TICER & RICER Multiple receivers in TICER
Performance analysis of TICER & RICER Multiple receivers in RICER
Performance analysis of TICER & RICER Multiple potential receivers As number of potential receivers –power consumption –Less sensitive to T It is beneficial when combined with routing
Conclusions Overview of rendezvous schemes in WSN Choose the Pseudo-Asynchronous scheme (TICER/RICER) Present the analysis in the presence of fading RICER performs better than TICER under strong fading conditions Be able to reduce power consumption substantially if the wakeup period T is chosen optimally The effect of channel fading on the performance of rendezvous schemes is major