Coded Task Offloading Boosts Privacy in Fluid Computing D2D Networks
New scheme uses secret sharing to balance speed, energy, and privacy in distributed systems.
Fluid computing aims to run distributed applications across heterogeneous cloud, edge, and device resources—but current task offloading rarely addresses privacy risks or energy awareness. A new paper from the University of Vigo tackles this with a coded task offloading scheme for D2D networks under stochastic task arrivals. The proposal integrates linear secret sharing (Shamir's scheme) where each task is split into coded shares, distributed to multiple devices. This redundancy supports threshold-based recovery (only a subset of shares needed), mitigates stragglers, and prevents any single device from seeing the full task, thus preserving privacy from adversaries.
The researchers formulate a privacy-aware optimization problem that jointly minimizes delay and energy consumption while penalizing theoretical privacy leakage (measured via mutual information under noisy observations). They solve it with a branch-and-bound solver and a lightweight heuristic scheduler, evaluated via discrete-event simulation. Compared to classic full offloading and parallel offloading, the coded scheme achieves better delay-energy Pareto frontiers. The heuristic runs near-optimality while being practical for real-time use. Importantly, results show that stronger privacy penalties shift offloading decisions, exposing a fundamental delay–energy–privacy trade-off. This work provides a principled foundation for privacy-preserving distributed computing in next-gen IoT and mobile edge networks.
- Combines linear secret sharing (Shamir’s scheme) with coded task offloading for threshold-based recovery and privacy preservation.
- Joint optimization of delay and energy with a privacy leakage penalty, solved via a branch-and-bound solver and a lightweight heuristic.
- Simulation shows coded offloading outperforms classical full/parallel offloading, and heuristic achieves near-optimal results, revealing a delay–energy–privacy trade-off.
Why It Matters
Enables secure, efficient distributed computing across heterogeneous devices, critical for privacy-sensitive IoT and edge applications.