2020/03/31 by Antoine Paris, Paris, Antoine, Hamed Mirghasemi +5
Computer Science · #Advanced Neural Network Applications #Distributed #FOS: Computer and information sciences #FOS: Electrical engineering #IoT and Edge/Fog Computing #Parallel #Signal Processing (eess.SP) #Stochastic Gradient Optimization Techniques #and Cluster Computing (cs.DC) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2004.00113
openalex publication_date 2020/03/31 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
Motivated by applications such as on-device collaborative neural network\ninference, this work investigates edge-facilitated collaborative fog computing\n- in which edge-devices collaborate with each other and with the edge of the\nnetwork to complete a processing task - to augment the computing capabilities\nof individual edge-devices while optimizing the collaboration for\nenergy-efficiency. Collaborative computing is modeled using the Map-Reduce\ndistributed computing framework, consisting in two rounds of computations\nseparated by a communication phase. The computing load is optimally distributed\namong the edge-devices, taking into account their diversity in term of\ncomputing and communications capabilities. In addition, edge-devices local\nparameters such as CPU clock frequency and RF transmit power are also optimized\nfor energy-efficiency. The corresponding optimization problem can be shown to\nbe convex and optimality conditions can be obtained through Lagrange duality\ntheory. A waterfilling-like interpretation for the size of the computing load\nassigned to each edge-device is given. Numerical experiments demonstrate the\nbenefits of the proposed optimal collaborative-computing scheme over various\nother schemes in several respects. Most notably, the proposed scheme exhibits\nincreased probability of successfully dealing with heavier computations and/or\nsmaller latency along with energy-efficiency gains of up to two orders of\nmagnitude. Both improvements come from the scheme ability to optimally leverage\nedge-devices diversity.\n