Imperial College

Research Group

Distributed Signal Processing in Sensor Networks

The paradigm of point-to-point communications is well established, and dates back to the seminal work of Shannon. With the recent advances of the network technology, a new paradigm for signal processing and communication is emerging and this will  have a dramatic impact on the way we acquire and process signals and the way we transport and reconstruct them. In our vision, multi-sensor networks are given by a large number of low-power, inexpensive, smart devices with multiple on-board sensors, connected through wireless links and the internet. The locations of the sensors may be known or unknown and their number may vary with time as sensors die or more are added. The realization of the potential of sensor networks requires the identification and solution of many new fundamental problems. In particular, since sensors are low-power devices and communication is critical, there are new trade-offs between accuracy of acquisition, delay,  computational and transmission power that need to be investigated. The main problems we plan to address include: understanding of distributed data acquisition schemes and structures, distributed versus centralized processing, fusion of the distributed sensed data to infer the original physical phenomenon, joint vs separate source-channel coding.

Main Publications:

Collaborations and Interactions:  M. Vetterli (EPFL),  M. Gastpar (UC, Berkeley) and  K.  Ramchandran  (UC, Berkeley) 

Department Home