Information Processing in Wireless Sensor Networks

Wireless sensor networks (WSNs) are formed via self-organization by a large number of micro-sensor nodes with low cost, low energy consumption and capability of sensing, signal processing and storing, and wireless communication. WSNs have drawn significant attention from the academic circles, military departments and industries for their wide potential applications. Realizing the potential of WSNs presents a number of challenges. Signal and information processing is one of the key technologies in it.

A number of research efforts are currently under way to address the issues surrounding signal and information processing in wireless sensor networks, including the methodologies for micro databases to collect, store, process data in the sensor network; methods to compile and execute queries and tasks; and mechanisms to deliver the results to end users, who may be mobile. One of the most critical areas is the distributed processing of the data collected from the devices, which is the focus of our research.

Focuses

Signal and information processing in distributed micro sensor networks is an emerging interdisciplinary research area, drawing upon traditionally disparate disciplines such as lower power communication and computation, space-time signal processing, distributed and fault-tolerant algorithms, adaptive systems, and sensor fusion and decision theory. Signal and information processing research into micro sensor networks has focused on developing new methods and algorithms for representing, storing, and processing spatially distributed, multimodal information.

In addition to considerations of single-platform signal processing, networked information processing is further constrained by application requirements on energy efficiency, network latency, and fault tolerance. With these factors in mind, the primary research focuses may be summarized as follows:

1. Intra-node collaboration: including multiple sensing modalities (combining acoustic and seismic measurements). No communication burden since collaboration is at a particular node, but higher computational burden at the node.
2. Inter-node collaboration: Combining measurements at different nodes. Higher communication burden since data is exchanged between nodes, and higher computational burden at fusion node.
3. Data fusion: Time series for different measurements are combined. Higher computational burden since higher dimensional data is jointly processed, and higher communication burden if different measurements from different nodes.
4. Decision fusion: Decisions (hard or soft) based on different measurements are combined. Lower computational burden since lower dimensional data (decisions) is jointly processed and higher communication burden if the component decisions are made at different nodes.

Future Directions

• Self organization
• Fault tolerance
• Constant False Alarm Rate (CFAR) event detection
• Target classification
• Target tracking