Heterogeneous Mobile Edge Computing
Lead Research Organisation:
University of Bristol
Department Name: Electrical and Electronic Engineering
Abstract
During the last five years, the global use of mobile devices such as smartphones and tablets has
increased greatly. According to projections made by Cisco, many people will have at least 7 devices
by the year 2020 and demand for high-quality mobile services on a constant basis is increasing. As a
result of these increased demands, mobile networks are experiencing challenges such as high load
and high bandwidth utilisation, further complicated by varying wireless access efficiency.
Various vendors and associations have striven to come up with new technologies. Among these
innovative proposals, the European Telecommunications Standards Institute (ETSI) developed the
concept of Mobile Edge Computing (MEC) to address the problem of the likely mismatch between
"traditional" cloud computing (with its finite number of data centres) and the increasing number of
mobile users. MEC is also recognized by the European 5G PPP as one of the key emerging
technologies for 5G networks (together with Software-Defined Networking (SDN)). Similar
terminologies and concepts have been introduced by Cisco ("Fog Computing") and "cloudlets" by
Carnegie Mellon University.
MEC proposes moving data processing capability away from distant consolidated data centres in the
cloud to servers that are physically located closer to the mobile user, in order to support higher
Quality of Experience. At the heart of MEC is the provision of computing power in a delocalized
manner to offer users the lowest possible latency, the highest possible bandwidth and direct access
to real-time network applications, such as gaming, video streaming etc. The distinguishing features of
MEC are its closeness to end-users, mobility support, and dense geographical deployment of the
MEC servers.
Despite the several advantages, realising the vision of MEC is a challenging task because of the
administrative policies. There is a need to investigate the potential opportunities and techniques for
enabling the vision of MEC. At the moment, research on MEC is at a very early stage, with many key
issues still open. As such, this PhD project will investigate and propose innovative solutions for some
of the following MEC-related topics:
* Smart MEC computation scheduling and offloading;
* Distributed data storage and caching
* MEC platform management
* Application-aware performance optimisation
* Radio network-aware content optimisation
* Real-time load prediction models
* Reliable and scalable resource allocation
* Context awareness
increased greatly. According to projections made by Cisco, many people will have at least 7 devices
by the year 2020 and demand for high-quality mobile services on a constant basis is increasing. As a
result of these increased demands, mobile networks are experiencing challenges such as high load
and high bandwidth utilisation, further complicated by varying wireless access efficiency.
Various vendors and associations have striven to come up with new technologies. Among these
innovative proposals, the European Telecommunications Standards Institute (ETSI) developed the
concept of Mobile Edge Computing (MEC) to address the problem of the likely mismatch between
"traditional" cloud computing (with its finite number of data centres) and the increasing number of
mobile users. MEC is also recognized by the European 5G PPP as one of the key emerging
technologies for 5G networks (together with Software-Defined Networking (SDN)). Similar
terminologies and concepts have been introduced by Cisco ("Fog Computing") and "cloudlets" by
Carnegie Mellon University.
MEC proposes moving data processing capability away from distant consolidated data centres in the
cloud to servers that are physically located closer to the mobile user, in order to support higher
Quality of Experience. At the heart of MEC is the provision of computing power in a delocalized
manner to offer users the lowest possible latency, the highest possible bandwidth and direct access
to real-time network applications, such as gaming, video streaming etc. The distinguishing features of
MEC are its closeness to end-users, mobility support, and dense geographical deployment of the
MEC servers.
Despite the several advantages, realising the vision of MEC is a challenging task because of the
administrative policies. There is a need to investigate the potential opportunities and techniques for
enabling the vision of MEC. At the moment, research on MEC is at a very early stage, with many key
issues still open. As such, this PhD project will investigate and propose innovative solutions for some
of the following MEC-related topics:
* Smart MEC computation scheduling and offloading;
* Distributed data storage and caching
* MEC platform management
* Application-aware performance optimisation
* Radio network-aware content optimisation
* Real-time load prediction models
* Reliable and scalable resource allocation
* Context awareness
People |
ORCID iD |
| Georgios Oikonomou (Primary Supervisor) | |
| Raghubir Singh (Student) |
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/P510427/1 | 01/10/2016 | 31/12/2021 | |||
| 1834874 | Studentship | EP/P510427/1 | 17/10/2016 | 16/10/2020 | Raghubir Singh |
| Description | The project has identified the key factors in how mobile device such as a smartphone can offload computationally complex tasks to Multi-Access Edge Computing (MEC) servers to achieve faster task completion time and greatly energy usage. The typical task for offloading include facial recognition and other real-time applications which can be completed quicker using the superior computing resources of MEC servers. Offloading also maximizes battery lifetime for mobile devices whose battery technologist have not kept pace with advances in mobile computational capacity. We have identified the speed increase offered by MEC servers, the complexity of the task, the communication speed between the mobile device and MEC network and the workloads of the processors as determining the success of offloading in a real-life situation where a mobile device with different computational abilities link via 4G network to MEC servers with widely variable workloads. Subsequently, I developed a novel mathematical approach to investigate factors that affect the success of offloading from mobile devices to Edge Computing servers. The most important conclusion was that very busy CPUs in both the mobile device and the server severely restricted the success of offloading. Additionally, the speed of network connection indicated that 5G networks would be very much better for offloading than 3G or 4G networks; 5G networks may run the risk of being overloaded with even low-complexity jobs from the increasing number of users of mobile devices. I then investigated how multiple jobs from a single mobile device could be offloaded in the local management system. I developed an allocation program in Pyomo/CPLEX. a popular optimization solver which could be run on a mobile device. This was validated by manual calculations by using Excel. The major conclusion was that a mobile device with a fast on-board processor would be very likely to partition jobs between local computation and offloading. Moving from a de-centralized approach to the centralized problem of resource allocation in Edge Computing network where servers communicate with multiple mobile devices, I devised an approach to asses the offloading options where many millions of possible schedules occurred. Most recently, I have begun analyzing how the user of a mobile device can optimize computation time, local energy usage and cost at the same time. |
| Exploitation Route | TRL aim to develop commercial MEC services. This is an area which has received much attention from Telecommunication and IT companies. |
| Sectors | Digital/Communication/Information Technologies (including Software),Healthcare,Transport |
| Description | All outcomes from the project have been shared with the Toshiba Research labs in Bristol to contribute to their technical development of Edge Computing technologies. The project is focused on the implementation of services related to Multi-Access Edge Computing and complements Toshiba's project such as the Park Us app real-time parking availability service using smartphone communication. Toshiba will be also able to use any program and algorithm developed in the course of this work |
| First Year Of Impact | 2020 |
| Sector | Digital/Communication/Information Technologies (including Software),Healthcare,Transport |
| Impact Types | Societal,Economic |