Latency Aware Resource Management At The Edge PDF Download

Efficient utilization of computing resources has always been an important challenge for service providers, leading to significant efforts on developing solutions, either in the form of new technology or new ways to enhance the efficiency of existing technologies. Mobile Edge Computing (MEC) is the latest technology developed to improve the high latency in mobile cloud computing systems which stems from the long distance between cloud servers and the end user. MEC systems are expected to improve the Quality of Service (QoS) by bringing servers closer to the end user, but when it comes to the cost of services, these systems face important challenges. The operating cost of MEC systems is higher than that of the remote clouds, due to the small servers which are distributed across the network. On the other hand, compared to the cloud data centers, edge nodes have more restricted capacity. Another challenge in MEC systems is the mobility of users, that might make the current allocation of resources inefficient or even infeasible in few minutes. These issues become more challenging in the Vehicular Edge Computing (VEC) systems where each vehicle can be considered as an edge node. In this dissertation, we address the mentioned challenges of resource allocation in MEC systems and VEC systems by designing efficient algorithms for resource management with the aim of improving the performance of these systems (i.e., energy consumption, operating cost, latency, and reliability). We address the Multi-Component Application Placement Problem (MCAPP) in MEC systems. We formulate this problem as a Mixed Integer Non-Linear Program (MINLP) with the objective of minimizing the total cost of running the applications. In our formula- tion, we take into account two important and challenging characteristics of MEC systems, the mobility of users and the network capabilities. We analyze the complexity of MCAPP and prove that it is N P -hard, that is, finding the optimal solution in reasonable amount of time is infeasible. We design two algorithms, one based on matching and local search and one based on a greedy approach, and evaluate their performance by conducting an extensive experimental analysis driven by two types of user mobility models, real-life mobility traces and random-walk. The results show that the proposed algorithms obtain near-optimal solutions and require small execution times for reasonably large problem instances.We also address the resource allocation and monetization challenges in MEC systems, where users have heterogeneous demands and compete for high quality services. We formulate the Edge Resource Allocation Problem (ERAP) as a Mixed-Integer Linear Pro- gram (MILP) and prove that ERAP is NP-hard. To solve the problem efficiently, we pro- pose two resource allocation mechanisms. First, we develop an auction-based mechanism and prove that the proposed mechanism is individually-rational and produces envy-free al- locations. We also propose an LP-based approximation mechanism that does not guarantee envy-freeness, but it provides solutions that are guaranteed to be within a given distance from the optimal solution. We evaluate the performance of the proposed mechanisms by conducting an extensive experimental analysis on ERAP instances of various sizes. We use the optimal solutions obtained by solving the MILP model using a commercial solver as benchmarks to evaluate the quality of solutions. Our analysis shows that the proposed mechanisms obtain near optimal solutions for fairly large size instances of the problem in a reasonable amount of time. Another contribution is VECMAN, a framework for energy-aware resource management in VEC systems. The main motivation behind VECMAN is to is improve the energy efficiency through sharing computing resources among connected EVs. However, the un- certainties in the future location of vehicles make it hard to decide which vehicles participate in resource sharing and how long they share their resources so that all participants benefit from resource sharing. VECMAN is composed of two algorithms: (i) a resource selector algorithm that determines the participating vehicles and the duration of resource sharing period; and (ii) an energy manager algorithm that manages computing resources of the participating vehicles with the aim of minimizing the computational energy consumption. We evaluate the proposed algorithms and show that they considerably reduce the vehicles0́9 computational energy consumption compared to the state-of-the-art baselines.