Fast Near-Optimal Result Deduplication and Relocation in Collaborative and Distributed Edge Computing Networks

The increasing complexity of collaborative and distributed edge computing networks necessitates efficient management of result caching and deduplication to optimize storage and reduce redundant processing. Traditional caching strategies often suffer from storage inefficiencies and computational overhead due to excessive result duplication across edge servers. To address this challenge, we propose a convex optimization framework for result deduplication and relocation (COFRDR) that strategically reallocates cached results while maintaining service quality. Unlike conventional heuristics, our approach transforms the inherently NP-hard result deduplication and relocation problem into a continuous convex optimization framework, allowing for faster convergence and better scalability across large-scale collaborative and distributed edge computing networks (CDECN). We introduce a fairness-aware storage balancing mechanism and a latency-space benefit (LSB) model to ensure optimal storage utilization across edge servers. A rounding algorithm is further integrated to map continuous solutions back to the binary domain, ensuring practical deployability. Extensive simulations confirm the effectiveness of our method, demonstrating significant improvements in storage efficiency, deduplication ratio (DR), and latency reduction compared to existing heuristics. The proposed framework offers a scalable and adaptive solution for future large-scale edge computing infrastructures, meeting the increasing demand for real-time, low-latency processing in IoT-driven environments.