Dynamic Edge User Allocation with User Specified QoS Preferences
Mobile Edge Computing (MEC) policies that bind user service requests to edge servers, seldom take into account user preferences of Quality-of-Service (QoS) and the resulting Quality-of-Experience (QoE). In this paper, we design a novel user-centric optima
- PDF / 433,955 Bytes
- 11 Pages / 439.37 x 666.142 pts Page_size
- 71 Downloads / 240 Views
Abstract. Mobile Edge Computing (MEC) policies that bind user service requests to edge servers, seldom take into account user preferences of Quality-of-Service (QoS) and the resulting Quality-of-Experience (QoE). In this paper, we design a novel user-centric optimal allocation policy considering the QoS preferences of users, with an attempt to maximize the overall QoE. Additionally, we propose a real-time mobility aware user-centric heuristic algorithm to solve the allocation problem by accommodating the time varying QoS demands of users. Experimental results on real data sets demonstrate the efficiency of our allocation scheme and a comparison with state-of-art approaches in MEC literature. Keywords: Edge computing
1
· Server allocation · User migration
Introduction
In recent times, Mobile Edge Computing (MEC) [1] has emerged as a new paradigm that allows service providers to deploy services on MEC servers located near base stations. As users move around, their application service invocations are routed to proximate MEC servers to curtail the high latencies of cloud communication networks. A service allocation policy is designed to determine the user-service-server binding, i.e. which service requests from which users are provisioned by which MEC servers in their vicinity, as they move around. In recent years, several allocation policies, static and dynamic, considering different optimization metrics have been proposed in literature [3,4,6–8]. The general philosophy of service allocation policies is to design and optimize a user-mobility aware service-server-user binding that optimizes some quantitative metric (e.g.. latency, energy, throughput) to cater to user application service needs and ensure seamless usage experience. A recent work [6] has proposed a novel view of considering qualitative QoS level offerings by service providers in designing the service bindings. Additionally, the authors have quantitatively correlated QoS values with overall Quality-of-Experience (QoE) of users to demonstrate the existence of thresholds, beyond which, enhancing QoS values no longer enhances a user QoE. This work, however, does not consider a user’s QoS preferences when deciding these bindings. Moreover, the binding is static, in other c Springer Nature Switzerland AG 2020 E. Kafeza et al. (Eds.): ICSOC 2020, LNCS 12571, pp. 187–197, 2020. https://doi.org/10.1007/978-3-030-65310-1_15
188
S. P. Panda et al.
words, once an allocation is decided for a user service invocation to a specific QoS level at an edge server, he is continued to be served at the same level throughout, oblivious to the fact that the user may not be in a position to enjoy services at a higher QoS level always due to battery or other constraints. Also, the policy is not adaptive, in the sense that user movements, joining or leaving of users, and user QoS preferences and preference changes in terms of the required QoS levels, are not accounted for. This motivated us to design a dynamic self-adaptive allocation policy that can address these variations. D
Data Loading...
