Mobile edge computing (MEC), also known as multi-access edge computing, is a networking technology that allows for the deployment of cloud-computing capabilities and an IT service environment at the edge of the network. As mobile networks have evolved over the years, MEC offers unique technical advantages compared to traditional cloud computing models by enabling the provisioning of IT and cloud-computing capabilities closer to mobile subscribers and devices. MEC has become an active area of research, given its advantages of low latency, location awareness, and wide-scale applications for 5G and Internet of Things (IoT) technologies. This article provides an in-depth review of recent MEC research papers focusing on key technical challenges and innovations related to optimization, resource management, network virtualization, security and privacy, applications, and testing frameworks.
One of the core challenges addressed in MEC research is optimization and resource management. A 2020 paper from researchers at Huawei proposed a deep reinforcement learning (DRL) method for optimizing the deployment of MEC servers and workload offloading in a multiserver environment. The DRL algorithm focused on minimizing service delay while considering factors like bandwidth utilization, computational resource usage, and deployment costs. Evaluation results showed the DRL approach could achieve significantly lower latency compared to other heuristics while keeping resource usage and deployment costs low. Another paper from 2021 examined computational resource optimization for MEC in an IoT environment using a Lyapunov optimization technique. The proposed algorithm jointly optimized CPU frequency adjustment and task scheduling to minimize energy consumption while satisfying latency constraints. Simulations validated that the approach outperformed standard methods in balancing energy-delay tradeoffs.
Network virtualization is another active area for MEC research given its ability to enable multi-tenancy, service flexibility, and efficient resource sharing. A 2020 paper proposed a network virtualization architecture for MEC combining network slicing, service function chaining, and network functions virtualization. The architecture supported isolation, programmability, and on-demand provisioning of virtualized MEC applications and infrastructure as a service. Performance evaluation showed the virtualization platform enabled efficient resource sharing while meeting latency and throughput requirements for diverse MEC applications. Security researchers have also investigated network virtualization techniques for MEC. A 2021 paper applied hardware-assisted virtualization to isolate MEC applications and infrastructure on shared edge devices. Experimental results verified the virtualization approach could prevent various software vulnerabilities and side-channel attacks between tenants with minimal performance overhead.
As MEC networks become more complex with diverse tenants, applications and device types, effective management plane architectures are crucial. A survey paper from 2021 evaluated various control and orchestration frameworks proposed for MEC including ETSI MEC, OSM, ONAP, and open-source initiatives. The comparison identified key functional requirements for scalable MEC management including device and service lifecycle management, application deployment automation, mobility support, and federated control across multiple administrative domains. The paper concluded that integrated management and orchestration will be critical to enable flexible and intelligent MEC services spanning wide-area networks. Another 2021 paper proposed a software-defined networking (SDN) architecture for joint MEC-mobile network control. The approach applied SDN principles for unified management of radio, transport and MEC compute resources. Testbed evaluation showed the SDN control plane could dynamically configure VNF placement, route traffic flows, and reconfigure resources in response to changing network conditions and application demands with low latency.
Security and privacy are major concerns in open MEC deployments due to the sensitive location and usage data processed at edge infrastructures. A survey from 2021 analyzed common MEC security threats and challenges including infrastructure attacks, side channels, software vulnerabilities and data leakage. The paper identified key requirements for future MEC security frameworks including distributed trust models, hardware-based isolation, anonymity techniques, and end-to-end encryption for application data and control planes. Researchers have also explored new techniques to enhance MEC privacy. A 2021 paper applied federated learning to a vehicle trajectory prediction application in MEC. By distributing model training across edge devices while keeping raw data localized, the approach significantly improved location privacy compared to central server training with much lower communication overhead. Overall, the dynamic multi-stakeholder environment of MEC demands innovative security architectures combining virtualization, cryptography and machine learning to protect users and applications.
With advancements in MEC infrastructure and management, research has expanded to prototype diverse latency-critical applications. A 2020 paper implemented high-definition map streaming using MEC to enable low-latency augmented reality navigation for autonomous vehicles. Edge servers preprocessed and cached map tiles based on vehicle routes and sensor data to reduce cloud query latency by 10x during testing. Researchers have also explored MEC for ultra-reliable low-latency industrial control through 5G dual connectivity. A 2021 prototype applied MEC to provide critical machine monitoring and actuation services with sub-10 millisecond latency for factory robots, wind turbines and oil rig equipment. Other promising MEC applications under investigation include emergency response systems using edge video analytics, tactile internet for teleoperation, and edge processing for intelligent transportation.
As MEC transitions from research to early commercial adoption, testing and validation of systems become increasingly important. A survey from 2021 analyzed existing MEC testbeds and their capabilities for experimenting with applications, algorithms, infrastructure and management planes. The paper identified gaps in mobility and interconnected testing environments to fully validate MEC capabilities. Researchers have since developed novel testbed frameworks to address such gaps. For example, a 2022 paper presented Olympus, a large-scale multi-access edge testbed comprising LTE/5G network slices, edge servers and emulated IoT devices. The framework supported testing of edge computing use cases across mobility, multi-tenancy, and wide-area connectivity. Ongoing work also aims to federate distributed MEC testbeds for continuous integration and end-to-end validation at scale. Standardized benchmarks are also being developed under the European Telecommunications Standards Institute (ETSI) to characterize performance of MEC platforms, servers and applications. Overall, open testbeds and tools will play a key role in advancing MEC technologies from research into mature commercial systems.
Mobile edge computing delivers compelling advantages over traditional cloud models through proximity, ultra-low latency and mobility support. This article reviewed recent advancements in MEC optimization techniques, network virtualization, management architectures, security & privacy solutions, prototype applications and testbed frameworks. Key research trends include application of machine learning, distributed computing paradigms, hardware security extensions and novel edge services for domains like industrial IoT. As network capabilities and computing resources at the edge continue expanding with 5G rollouts, MEC promises to transform a wide range of latency-critical domains in the coming years. Ongoing standardization efforts and open testbeds will accelerate adoption and help realize the full potential of edge computing technologies across industries.
Andrew Stroud