Collaborative Computing Strategy Based SINS Prediction for Emergency UAVs Network

TL;DR

This paper proposes a collaborative computing strategy with SINS prediction for emergency UAV networks, improving latency and task success rate under dynamic conditions through a novel graph-based task scheduling and optimization approach.

Contribution

It introduces a two-step weighted time expanded graph for dynamic topology handling and employs BPSO for optimal task mapping, enhancing UAV network performance in emergencies.

Findings

Significant latency reduction compared to cloud and local computing.

Improved task success rate with SINS-based prediction.

Effective handling of dynamic network topology changes.

Abstract

In emergency scenarios, the dynamic and harsh conditions necessitate timely trajectory adjustments for drones, leading to highly dynamic network topologies and potential task failures. To address these challenges, a collaborative computing strategy based strapdown inertial navigation system (SINS) prediction for emergency UAVs network (EUN) is proposed, where a two-step weighted time expanded graph (WTEG) is constructed to deal with dynamic network topology changes. Furthermore, the task scheduling is formulated as a Directed Acyclic Graph (DAG) to WTEG mapping problem to achieve collaborative computing while transmitting among UAVs. Finally, the binary particle swarm optimization (BPSO) algorithm is employed to choose the mapping strategy that minimizes end-to-end processing latency. The simulation results validate that the collaborative computing strategy significantly outperforms both cloud and local computing in terms of latency. Moreover, the task success rate using SINS is substantially improved compared to approaches without prior prediction.