On the Robustness of Distributed Computing Networks

Abstract

Traffic flows in a distributed computing network require both transmission and processing, and can be interdicted by removing either communication or computation resources. We study the robustness of a distributed computing network under the failures of communication links and computation nodes. We define cut metrics that measure the connectivity, and show a non-zero gap between the maximum flow and the minimum cut. Moreover, we study a network flow interdiction problem that minimizes the maximum flow by removing communication and computation resources within a given budget. We develop mathematical programs to compute the optimal interdiction, and polynomial-time approximation algorithms that achieve nearoptimal interdiction in simulation.

Minimization of Energy and Service Latency Computation Offloading using Neural Network in 5G NOMA System

“…A joint radio communication, caching, and computing decision problem is developed to maximize the average tolerant delay while satisfying a specified transmission rate constraint to be able to optimize resource allocation at both mobile VR devices and fog access points (F-APs) [13,14]. In a centralized computing network, traffic flows should be transmitted and processed, and they can be stopped by cutting off the resources needed for either communication or computation [15,16]. Using the caching and cooperative communication abilities of the terrestrial Base Stations (BSs) and unmanned aerial vehicles (UAVs), a distributed heterogeneous computing platform (HCP) is created.…”

Smart Grid Information Processes Using IoT and Big Data with Cloud and Edge Computing

“…Another benefit of using edge computing in SG is the reduction in failure-if there is an electricity outage problem in a particular area of the grid, the edge computing services of the other areas will operate normally, without any problem. On the other hand, if the grid relies solely on cloud computing, and there is a power supply failure due to any natural disaster in the cloud infrastructure, then the whole network will fail [27]. As shown in the Figure 4, cloud computing shows the best performance when the signal to noise ratio (SNR) is low, but the edge computing performs the best even at high-SNR regime as the number of edge servers increases, outperforming the cloud-assisted counterpart.…”

Key Advances in Pervasive Edge Computing for Industrial Internet of Things in 5G and Beyond

“…For instance, if an electricity outage happens in a particular area of the grid, the edge computing services of other areas will continue to operate normally, unaffected. In contrast, if a given IIoT network relies solely on centralized cloud computing, when the electricity supply fails due to any natural disaster happening in the cloud infrastructure, or the backhaul communication link becomes unstable, the whole network will fail [34], [35].…”

“…These results provide a clear example of the tolerance to failures of a PEC system. Nevertheless, despite the resiliency that PEC can provide to IIoT, systemic software failures can still happen, and this can result in a generalized network collapse, as reported in [34], [36], [37].…”