Target identification processor for wireless sensor network

Wireless sensor networks have many applications in the commercial world as well as in the military. In the area of defense applications, distributed Wireless sensor networks will provide new information systems for both military battlefield situational awareness and national security. In particular, the wireless sensor network can be used for target and threat identification in the battlefield environment. The objective of this thesis is to develop an architecture for a target recognition processor that can be implemented readily using on-node microprocessors of the sensor network. The architecture of the target recognition processor consists of three sub-processors: (1) source separation and deconvolution processor, (2) feature extraction processor, and (3) pattern recognition processor.; For the source separation and deconvolution processor, we consider a subspace based algorithm for blind source separation and deconvolution. A frequency domain parametric representation algorithm is chosen for the feature extraction processor. Finally, we analyze the performance of a Hidden Markov Model (HMM) based pattern recognition processor. Each subprocessor is considered and optimized in this research. In this thesis, we present a modular architecture for the target recognition processor to be implemented on a wireless sensor network. The modular structure simplifies the design optimization process at each stage versus the all-in-one approach. The proposed architecture further permits a set of choices of algorithms at deeper levels. With signal processing capabilities added to sensor networks, we are able to turn traditional information gathering sensor networks into smart networks that provide knowledge and intelligence for the environment. These smart sensor networks will greatly simplify the knowledge and intelligence gathering process for the decision making. As a result, a new generation of efficient systems is made possible with the deployment of these sensor systems.