Intelligent distributed architecture (IDA) for mobile sensor data fusion

摘要

In this paper, we provide a new two-layer scalable architecture, intelligent distributed architecture (IDA). The first layer of IDA, the application layer, is responsible for the mission specific tasks. The second layer, the network layer, is responsible for relaying the information quickly to reduce latency. In order to increase efficiency, we propose an intelligent network layer that evaluates the network traffic through application provided policies. The evaluation assesses improved accuracy in bandwidth limited scenarios. IDA policies permit the application to drop, merge or modify the packets in real time from different sources. We demonstrate how this architecture can be applied to an example application, distributed target tracking. We also propose some new algorithms that can be used in conjunction with IDA for target tracking. Our experiments on this application show that IDA improves system performance when the bandwidth is limited.