Semantic Information Fusion for Coordinated Signal Processing in Mobile Sensor Networks

Abstract

Distributed cognition of dynamic processes is commonly observed in mobile groups of animates like schools of fish, hunting lions, or in human teams for sports or military maneuvers. This paper presents methods for dynamic distributed cognition using an ad hoc mobile network of microsensors to detect, identify and track targets in noisy environments. We develop off-line algorithms for aggregating the most appropriate knowledge abstractions into semantic information, which is then used for on-line fusion of relevant attributes observed by local clusters in the sensor network.Local analysis of time series of sensor data yields aggregated semantic information, which is exchanged across nodes for higher level distributed cognition. This eliminates the need for exchanging high volumes of signal data and, thus reduces bandwidth and energy requirements for battery powered microsensors.

Distributed target classification and tracking in sensor networks

“…In the numerical results presented below, typically one CPA event from each of three modalities of four to five nodes was used in this calculation (12 to 15 total). The results in [13] show this to be a reliable technique.…”

On Topology of Sensor Networks Deployed for Multitarget Tracking

“…Various advanced techniques have been applied to solve the problem. Friedlander and Phoha [8] proposed statistical techniques based on principal component analysis for target identification with knowledge on target position, time, target velocity, and a set of target attributes collected from sensors. In [9], the authors proposed wavelet-based algorithms for target classification.…”

PhotoNet: A Similarity-Aware Picture Delivery Service for Situation Awareness

“…Semantic fusion has usually two phases: knowledge base construction and pattern matching [19]. At first phase, a suitable abstraction for representing semantic information is chosen, which is then used in second phase for matching and fusing relevant attributes.…”

“…This fusion runs in-network inference processes so that nodes only exchange semantic interpretations. In [19], authors applied semantic fusion for target classification. Another work [20] integrates sensor data into formal languages, and then matches data with some stored knowledge base based on the hypothesis that data represented by similar languages are semantically similar.…”