Project Objectives:

The rapid increase in demand for mobile devices within the realm of augmented reality and gaming applications motivate the need for real-time mobile cloud services. Edge cloud computing moves cloud services to the edge of the network, thereby, allowing clients to access services with a significantly reduced network delay. This service migration is intended to enable a range of latency sensitive mobile applications. In this project, we propose to manage user QoS by actively migrating services to different edge clouds in response to degraded server or network performance.

Technology Rationale:

Previous studies have proposed a distance-based Markov Decision Process (MDP) for optimizing migration decisions. These models provide the feasibility of applying MDP to edge cloud service migration decisions. However, these models fail to consider dynamic network and server states in migration decisions. In this work, we address these limitations by designing a comprehensive edge cloud migration decision system, which we call SEGUE. SEGUE achieves optimal migration decisions by providing a long-term optimal QoS to mobile users in the presence of link quality and server load variation.

Technical Approach:

The SEGUE system is designed to reliably answer those two critical migration questions of when and where. SEGUE is able to choose the optimal server for a mobile client's serve migration considering the network and server states. SEGUE adopts a QoS aware scheme to activate the MDP model when a QoS vio- lation is predicted to solve for the when to migrate variable. Two components of SEGUE work together to achieve this. A state collection module monitors and collects key parameters continuously, parameters representing network states, server workloads and user mobility. The other component of the SEGUE QoS aware scheme to determine the When to challenge is a QoS prediction mod- ule. The QoS prediction module uses state collection module inputs to predict QoS. When a possible QoS violation is detected, the QoS prediction module compels the SEGUE system to run the state based MDP which then selects a target server through the edge cloud selection module, accordingly. This al- lows SEGUE to avoid unnecessary migration costs and bypass any possible QoS violations. Once a best Edge Cloud is selected, the service migration module migrates the service from the source edge cloud to the optimal edge cloud.

Results To Date and Future Work Plan:

We evaluate SEGUE performance using an augmented reality application applied to a real world scenario. The augmented reality application is a representative mobile application that has the two-fold stringent criterion of low latency and high bandwidth. Our evaluation's reliability is assured through the real mobility trace of 320 taxis in Rome. Our evaluation itself is realized through an end-to-end response time of the augmented application and workload distribution among edge clouds. We also compare the performance of SEGUE with two other edge cloud service migration decision models: lowest load edge cloud migration, and least hop edge cloud migration, which are designed to provide optimal QoS when to specifically consider the server load states and the network states. The average response time provided by SEGUE, the lowest workload migration and the least hop migration: 224.04ms, 307.80ms and 483.86ms, respectively. SEGUE improves QoS by 27.21% and 53.70% at this condition. SEGUE recognizes, by design, that both the edge cloud server and network states are dynamic. Exclusive consideration of one state invariably fails to maintain the optimal QoS in the other state, both of which comprise two in- dependent dynamic real-time states. SEGUE proves its adaptive nature as it smoothly integrate both dynamic server and the network states into one best- case migration solution, thereby maintaining the optimal QoS of edge cloud service. We believe that the proposed technique will enable real-time cloud applications in future mobile environments. Future work will consider experimental proof-of-concept validation of the simulation based results provided here and also evaluate the resource cost of this proposed system.