Sensing Algorithms for Bandwidth Efficient Edge Radars (SABER)

Distributed sensing radars serve as an important component of the Joint Force strategy for situational awareness, e.g., US Navy’s vision of Distributed Maritime Operations (DMO) [1] and advanced tactical imaging research. By networking physically dispersed radar nodes into a unified cooperative mesh, the Joint Force can fuse information to enhance surface target detection, track and identification. Furthermore, coordinating information from these dispersed nodes overcomes traditional single-platform limitations [1] such as single points of failure and limited angular diversity, especially in contested, high clutter environments. The transition to decentralized sensing and DMO fundamentally shifts the burden of performance from individual platform sensors to the underlying communication networks. Functioning as a cohesive, decentralized system requires synchronization and low-latency data links. This is especially critical for imaging nodes. The fidelity of a decentralized system is directly proportional to the amount of data the network can move. Streaming raw radio frequency (RF) data allows for maximum imaging resolution but severely strains network capacity. Conversely, enabling nodes to process their own radar data and only sharing images or features reduces bandwidth requirements but results in degraded imaging in collaborative scenarios. Consequently, there is an operational need for new architectures and processing frameworks that explicitly relate communication, computation and sensing resources to surface target detection, tracking and identification performance metrics. To address these challenges, this SBIR topic seeks solutions that jointly optimize communications and target inference in a decentralized radar sensing framework [2, 3, 4] that processes data locally and only transmits actionable intelligence or compressed imaging features into the network.