Application-Aware Resource Allocation and Data Management for MEC-assisted IoT Service Providers

TL;DR

This paper tackles joint IoT service placement, data management, and resource allocation in MEC-enabled networks, explicitly accounting for data dependencies and cross-service data caching. It introduces an IoT Service Manager within the ETSI MEC framework to enable data sharing while preserving business data isolation, and formulates a nonlinear MINLP that is linearized to a MILP for tractable solutions. An LR-based heuristic provides near-optimal solutions with significantly lower runtime, validated against multiple benchmarks under nonuniform traffic patterns. The findings show substantial gains in admitted services and balanced resource utilization, demonstrating practical benefits for MEC-based IoT service providers in urban, data-intensive scenarios.

Abstract

To support the growing demand for data-intensive and low-latency IoT applications, Multi-Access Edge Computing (MEC) is emerging as an effective edge-computing approach enabling the execution of delay-sensitive processing tasks close to end-users. However, most of the existing works on resource allocation and service placement in MEC systems overlook the unique characteristics of new IoT use cases. For instance, many IoT applications require the periodic execution of computing tasks on real-time data streams that originate from devices dispersed over a wide area. Thus, users requesting IoT services are typically distant from the data producers. To fill this gap, the contribution of this work is two-fold. Firstly, we propose a MEC-compliant architectural solution to support the operation of multiple IoT service providers over a common MEC platform deployment, which enables the steering and shaping of IoT data transport within the platform. Secondly, we model the problem of service placement and data management in the proposed MEC-based solution taking into account the dependencies at the data level between IoT services and sensing resources. Our model also considers that caches can be deployed on MEC hosts, to allow the sharing of the same data between different IoT services with overlapping geographical scope, and provides support for IoT services with heterogeneous QoS requirements, such as different frequencies of periodic task execution. Due to the complexity of the optimisation problem, a heuristic algorithm is proposed using linear relaxation and rounding techniques. Extensive simulation results demonstrate the efficiency of the proposed approach, especially when traffic demands generated by the service requests are not uniform.

Figures (11)

• Figure 1: Illustration of the high-level system architecture. Different colours represent different types of IoT services. Dashed coloured lines represent IoT data traffic. The figure shows that three services belonging to two service types are deployed in the MEC system.
• Figure 2: (Simplified) MEC reference architecture and functional components.
• Figure 3: Functional architecture of the IoT Service Manager.
• Figure 4: Evaluation setup
• Figure 5: Percentage of deployed IoT services versus the number of service requests for uniform cell popularity (Scen A).