PIM: A unique framework for sensor fusion in the tactical environment

Abstract

Tactical environments are usually scattered with a number of sensors that collect different types of information. This data can be correlated, combined, and otherwise processed using sensor fusion algorithms to provide warfighters with a more accurate representation of the situation and environment. Sensor fusion algorithms can dramatically increase situational awareness; to accomplish this, they require access to the data collected by multiple sensors on a fusion node. In such an approach, the amount of data collected by the sensors and sent over the network can be very large: some sensors may generate an update every few milliseconds; others may generate large updates such as imagery; a substantial number of sensors may be present in the network. Moving large amounts of data in a tactical environment is a complex task due to the low bandwidth available and the unreliability of the channel typical of the wireless networks in the field. We aim to improve the bandwidth utilization to increase the sensor fusion algorithm performance and propose to achieve this goal using the novel concept of a Process Integrated Mechanism (PIM): instead of sending the information to a centralized node for processing, the PIM framework transparently migrates the fusion algorithm that processes the information to the sensor nodes. Using the PIM middleware, only the current state of the fusion process is sent from one node to the next one, granting the process seamless access to the information as if it were stored locally. Our findings show that in a variety of cases the bandwidth consumption of our approach is lower than the bandwidth needed using other frameworks.