Energy-Efficient Data Gathering in Sensor and Actor Networks: A High Bit-rate Image Sensing Application Rajnish Kumar, Mina Sartipi, Junsuk Shin, Ramanuja Vedantham, Yujie Zhu, Faramarz Fekri, Umakishore Ramachandran, Raghupathy Sivakumar Application Distributed Source Coding of Correlated Data from Image Sensors Mutual Exclusion for Command Delivery from Base Station to LED Array Actors Heterogeneous wireless sensor and actor network consisting of mica2 motes with light sensors, LED array actors, IPAQs with image sensors (cameras), where Light sensors report the light readings periodically to the Base Station (BS) LED array actors are turned on based on the light readings Base station sends a command to cameras to turn on the camera after LED arrays are on Cameras send the image data to the BS Motivation: Motivation: Need for mutual exclusion in the acting ranges of the LED arrays Mutual Exclusion in WSANs: Execute a given command exactly once (or desired number of times) for any particular location irrespective of the distribution of actors Relaxed Definition: Choose a minimal set of actors such that the overlap between acting regions is minimal Goal: Minimize the total number of transmissions for the three phases for energy-efficient communication Illustration of Mutual Exclusion: Definitions for illustration Rm: Region covered by set of actors already included as part of actor cover ri and rj: New area covered by actor i and j respectively ni and nj: New overlap area for actor i and j respectively oi and oj : Old overlap area for actor i and j respectively Image sensors have correlated data. Distributed source coding can exploit correlation structure with low power algorithms Correlation Model: X1, X2 : I.I.D binary sequence; Prob [ Xi =0 ] = Prob [ Xi=1 ] = 1/2. Prob [ X1 ≠ X2 | X1 ] = p BSC X2 X1 p Distributed Source Coding: X1 and X2 have correlated information. Goal: Compressing X2: Without communicating with X1 With the knowledge that X1 is present at the decoder Modeling Distributed Source Coding with Parallel Channels: Non-uniform Channels Sensor Stack with Cross-layering support for efficient Image sensor placement Energy-efficient communication strategy from Light sensors to Base Station X2 k Motivation: Motivation: Need for energy-efficient communication from light sensors to sink Traditional communication strategy conveys information between the sender and the receiver using energy (EbT) only Energy consumption is keb, where k is the length of the bit-stream and eb is energy per bit Can we use time as an added dimension to convey information? Communication through Silence (CtS): A new communication strategy that conveys information using silent periods in tandem with small amount of energy The energy consumption for CtS is always 2eb irrespective of the amount of information being sent EbT 97 Cross-layering can help in better camera placement for the application considered Without cross-layering, there is information overlap across layers Modules make inefficient decision – DFuse application needs routing information to decide about role migration 1 Role Assignment 9 3 8 HSN Routing 6 Time Sync Service 7 MAC 0 0 0 1 97 ! 1 1 0 0 0 0 1 1 1 0 0 0 0 1 1 Information Exchange Service: 97 1. CtS START 1 2 …. 97 ! 2. Efficient use of limited memory Simple interface for information sharing Data Fusion Layer Data Service Layer In-stack fusion Information Exchange Service Extensibility Radio STOP 4. Next-hop selection, Logical naming, Packet scatter/gather AttributeValue publish/ subscribe Connection Asynchronous delivery (A) Stack Lay-out 5. Helper Service Layer Medium Access, Error Control, Radio Control Medium Access Layer 3. \ Logic Application Application Complex event notification (B) Functionalities PX2 (1-R)n Wireless Channel X2 P'X2 Future Work: Distributed source coding for image sensors Sensor Stack with Cross-layering support: 1 0 Decoder Conclusions: Energy savings for distributed source coding: 40% Energy savings for cross-layer support: 110% Energy savings for energy-efficient communication: 88% Energy savings for Mutual exclusion for LED array actors: 55% Overall expected energy savings: 88 + 55 + 110+ 40 = 293% 2 Flood Routing X1 Conclusions and Future Work Application, Data Fusion 4 5 Correlation Channel Use non-uniform LDPC code for channel coding. Sensor Stack without Cross-layering support: 1. Fusion Requirement 2. Data transmission requirement 3. Neighborhood 4. Data transmission requirement 5. Data transmission requirement 6. Neighborhood, Topology 7. Time synchronization accuracy requirement 8. Data transmission requirement 9. Role schedule, Duty cycle information Systematic (X2 ,PX2 ) Channel n Encoder Rate R X2 Rn Localization, Synchronization Service Implement the algorithm on image sensors to evaluate energy saving benefits Cross-layer support for image sensor placement Implement the IES architecture for the heterogeneous testbed for data fusion Energy-efficient communication from light sensors to the BS Implement CtS communication strategy from light sensors to the BS Mutual exclusion for LED array actors Implement mutual exclusion on LED arrays to minimize energy consumption