Energy-efficient Energy-efficient Data Collection in Wireless Data Collection in Wireless Sensor Networks Sensor Networks G. Anastasi Pervasive Computing & Networking Lab. (PerLab) Department of Information Engineering University of Pisa, Italy PerLab g.anastasi@iet.unipi.it WIRTES 2007 Pisa, July 2, 2007 Research Topics � Pervasive and Mobile Computing � Wireless Sensor Networks � Opportunistic Networking � Ad Hoc and Mesh Networks � Power management for mobile computing Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 1
Research Topics � Pervasive and Mobile Computing � Wireless Sensor Networks � Opportunistic Networking � Ad Hoc and Mesh Networks � Power management for mobile computing Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 Sensor Node Architecture A Sensor Node A Sensor Node Location Finding System Mobilizer Mobilizer Location Finding System � Small � Low power Processor Processor � Low bit rate Sensor ADC Transceiver Transceiver Memory Memory � Low cost Power Unit Power Unit Power Generator Power Generator Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 2
Dense Sensor Networks Sink Internet, Internet, Satellite, Satellite, Sink Sensor etc etc Nodes Field Sensor Sink Field Task Manager � Several thousand nodes � Nodes are tens of feet of each other � Densities as high as 20 nodes/m3 I.F.Akyildiz, W.Su, Y. Sankarasubramaniam, E. Cayirci, “Wireless Sensor Networks: A Survey” � Multi-hop communication Computer Networks (Elsevier) Journal , March 2002. Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 Sparse Sensor Networks � Distance between nodes larger than transmission range � Data collection through mobile nodes (data mules) � Part of the external environment (buses, cabs, …) � Part of the infrastructure (robots, …) Base Station R M S. Jain, R. Shah, W. Brunette, G. Borriello, S. Roy Data “Exploiting Mobility for Energy Efficient Data Collection in Wireless Sensor Networks” Mule ACM/Springer Mobile Networks and Applications , Vol. 11, pp. 327-339, 2006. Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 3
Data Collection through Data Mules � Pros � Increased system lifetime � Increased reliability � Increased capacity � Increased flexibility � Cons � Increased message latency (delay-tolerant applications) � Limited scalability (unless multiple mules are used) � Physical obstacles may limit mule’s movements � Costs of data mule(s) Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 How to prolong the network lifetime? � Dense Sensor Networks � Sparse Sensor Networks 4
How to prolong the network lifetime? � Dense Sensor Networks � Sparse Sensor Networks Possible approaches � Energy Harvesting � Low-Power Components and Design � Topology Management � Power Management Management � Data compression/aggregation � Optimal Sampling � Predictive Monitoring � … Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 5
Power Management Adaptive Duty Cycling Sleep/Wakeup Adaptive Scheduling sampling � Sleep Wakeup Scheduling � Switches off the radio subsystem during inactivity periods � Sleep/wakeup schedule needed for communication � Adaptive sampling � Reduces the amount of data to be transmitted to the sink � Decreases the power consumption for sensing Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 Adaptive Sleep/Wakeup � Goal: minimize radio energy consumption � Basic Idea of Sleep/Wakeup Schemes � Nodes sleep for most of the time � Wake up periodically for transmitting/receiving data � MAC or Application-layer protocol � Our Scheme � Application-layer protocol � Relies on a routing tree � Active periods dynamically adjusted Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 6
Multi-hop Network Routing Tree Data flow from leaves to the root (sink node) ~Communication Period ~ i Coordination scheme j .Talk Interval . .Silence Interval . Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 Fully Synch (TinyDB) 1 ... 2 3 ... 4 ... � Cons � Pros � Static scheme � Simplicity � Global duty-cycle (low efficiency) � Requires clock synchronization Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 7
Fixed Staggered (TAG) 1 ... 2 3 ... 4 ... � Parent-child talk intervals � Adjacent to reduce sleep-awake commutations � Clock synchronization (periodic timestamps sent by the sink node) � Cons � Pros � Static scheme � Global parameters � Pipeline � Suitable to data aggregation Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 Our Proposal: Adaptive Staggered 1 2 3 4 � Adaptive duty cycle � Variable-length talk intervals depending on � Number of children � Network traffic � Channel conditions � Node Joins/Leaves � … Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 8
Simulation � Simulation Setup � Ns-2 tool � IEEE 802.15.4 MAC protocol � Network scenario � 50m×50m area � 30 nodes randomly deployed � Performance Indices � Average Activity Time (in % wrt always on) � Delivery Ratio � Average Message Latency � Fairness (MRD) Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 Simulation Results � Adaptation example Sample topology Talk interval adaptation Initial phase Node Talk interval (secs) # Steady-state TX Range = 15m, CS Range = 30m, 50m x 50m grid, Communication period # 30 nodes Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 9
Simulation Results � Comparison with other approaches Global delivery ratio 84,00% 82,00% 80,00% 78,00% Delivery ratio 76,00% 74,00% 72,00% 70,00% 68,00% 66,00% Adaptive staggered Fixed staggered Fixed staggered Alw ays on (optimal) (TAG) End to end latency Protocol fairness 500 8,00% 450 7,00% 400 Mean relative deviation (MRD) 6,00% 350 Latency (msecs) 5,00% 300 4,00% 250 200 3,00% 150 2,00% 100 1,00% 50 0,00% Adaptive staggered Fixed staggered Fixed staggered Always on 0 (optimal) (TAG) Adaptive staggered Fixed staggered (optimal) Alw ays on Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 Power Management Adaptive Duty Cycling Sleep/Wakeup Adaptive sampling Scheduling � Sleep Wakeup Scheduling � Switches off the radio subsystem during inactivity periods � Sleep/wakeup schedule needed for communication � Adaptive sampling � Decreases the power consumption for sensing � Reduces the amount of data to be transmitted to the sink Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 10
Snow Sensor Courtesy by E. Pasero (PoliTo), C. Alippi (PoliMi) Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 An Embedded System to Evaluate the Snow Status ice water snow 45000 Snow 40000 Snow Snow 35000 Snow Equivalent capacity 30000 Snow Snow 25000 Ice Ice 20000 Ice 15000 Ice Ice 10000 Ice Ice 5000 Ice Air 0 Water 50 Hz 100 Hz 200 Hz 500 Hz 800 Hz 1 KHz 5 KHz 10 KHz 50 KHz 100 KHz Measuring frequency Courtesy by E. Pasero (PoliTo), C. Alippi (PoliMi) Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 11
Snow sensor Power consumption Power consumption Sensing only Sensing, processing and communication Courtesy by E. Pasero (PoliTo), C. Alippi (PoliMi) Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 Adaptive Sampling Algorithm Estimate F max by considering the initial W samples and set F c = c * F max .; Define F up = (1 + (c-2)/4) * F max and F downp = (1 – (c-2)/4) * F max ; h 1 =0 and h 2 =0; for (i=W+1; i < DataLength; i++) { Estimate the current maximum frequency F cur on the subsequence (i-W, W) if ( | F curr - F up | < | F curr - F max | ) h 1 = h 1 +1; else if ( | F curr - F down | < | F curr - F max | ) h 2 = h 2 +1; else { h 1 =0; h 2 = 0; } if ( h 1 > h )|| ( h 2 > h) { F c = c * F curr .; F up = (1 + (c-2)/4) * F curr ; F down = (1 – (c-2)/4) * F curr ; } } Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 12
Simulation � Simulation Setup � MatLab � Experimental datasets � Network Scenario � Star Topology � TDMA communication scheme � Adaptive sampling algorithm at Base Station � Perfomance Indices � % of samples wrt fixed over-sampling � Mean Relative Error (MRE) Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 Simulation Results � Sampling Fraction (17-26%) Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 13
Simulation Results � Graphical Comparison Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 Simulation Results � Energy Consumption Activity Power management scheme Power cons. ratio Always On 880 mJ/sample 100% Duty-cycle 150 mJ/sample 17% Duty-cycle + Adaptive 3-5% Sampling Energy-efficient Data Collection in Wireless Sensor Networks WIRTES 2007 14
Recommend
More recommend
