Uncoordinated Access to Serverless Computing in MEC Systems for IoT

TL;DR

This paper tackles fast-changing edge conditions for IoT offloading by introducing uncoordinated serverless access in MEC, where each client is given a pool of executors and makes lightweight, per-request decisions. It combines queuing-theoretic analysis using $M/M/1$ PS queues with a discrete-time Markov chain to model state transitions and long-run delays, and validates the approach with testbed emulations that compare against centralized, distributed, and static baselines. The key finding is that uncoordinated access can significantly reduce latency jitter and maintain competitive delays with far lower coordination and backhaul overhead, making it practical for large-scale, heterogeneous IoT deployments and compatible with ETSI MEC. The work demonstrates notable improvements in latency and network efficiency, supporting deployment in real MEC environments and outlining paths for future improvements such as closed-loop optimization and tighter orchestration integration.

Abstract

Edge computing is a promising solution to enable low-latency IoT applications, by shifting computation from remote data centers to local devices, less powerful but closer to the end user devices. However, this creates the challenge on how to best assign clients to edge nodes offering compute capabilities. So far, two antithetical architectures are proposed: centralized resource orchestration or distributed overlay. In this work we explore a third way, called uncoordinated access, which consists in letting every device exploring multiple opportunities, to opportunistically embrace the heterogeneity of network and load conditions towards diverse edge nodes. In particular, our contribution is intended for emerging serverless IoT applications, which do not have a state on the edge nodes executing tasks. We model the proposed system as a set of M/M/1 queues and show that it achieves a smaller jitter delay than single edge node allocation. Furthermore, we compare uncoordinated access with state-of-the-art centralized and distributed alternatives in testbed experiments under more realistic conditions. Based on the results, our proposed approach, which requires a tiny fraction of the complexity of the alternatives in both the device and network components, is very effective in using the network resources, while incurring only a small penalty in terms of increased compute load and high percentiles of delay.

Figures (19)

• Figure 1: Target scenario.
• Figure 2: Simplified blueprint of the ETSI MEC reference architecture.
• Figure 3: Comparison of the proposed architecture for uncoordinated access to serverless functions (bottom) with traditional centralized (top) and distributed (middle) approaches.
• Figure 4: Sequence diagram of the execution of lambda functions from a client towards its primary executor and, once every $\chi$ requests on average, to the secondary executor, as well, for probing purposes. In the figure $\delta$ is the overall delay, consisting of processing component $\eta$ and network component $\tau$; the full notation is introduced in Sec. \ref{['sec:model']} and summarized in Table \ref{['tab:notation']}.
• Figure 5: System model, illustrated by means of an example with two clients and two executors, with $\mathbf{\tau} = \mathbf{x} = \mathbf{0}$; the equivalent M/M/1 model is on the right.