Distributed genetic algorithm for application placement in the compute continuum leveraging infrastructure nodes for optimization

TL;DR

This paper tackles resource optimization in fog computing by proposing three in-situ, distributed NSGA-II designs (Semi-distributed, Fully-distributed, and Neighbor-aware) that run on fog devices with a cloud coordinator. Using the fog application placement problem (FAPP) as a benchmark, the authors evaluate the trade-offs between solution quality and network overhead, comparing against a centralized NSGA-II baseline. The Semi-distributed design closely matches centralized performance but incurs higher network load, while Fully-distributed reduces network traffic at a cost to diversity and objective richness; the Neighbor-aware design minimizes communication yet delivers the weakest optimization performance. The work provides practical guidance for deploying distributed GAs in fog environments and outlines future directions, including hybridization and improved sub-population metrics to boost performance with low overhead.

Abstract

The increasing complexity of fog computing environments calls for efficient resource optimization techniques. In this paper, we propose and evaluate three distributed designs of a genetic algorithm (GA) for resource optimization in fog computing, within an increasing degree of distribution. The designs leverage the execution of the GA in the fog devices themselves by dealing with the specific features of this domain: constrained resources and widely geographical distribution of the devices. For their evaluation, we implemented a benchmark case using the NSGA-II for the specific problem of optimizing the fog service placement, according to the guidelines of our three distributed designs. These three experimental scenarios were compared with a control case, a traditional centralized version of this GA algorithm, considering solution quality and network overhead. The results show that the design with the lowest distribution degree, which keeps centralized storage of the objective space, achieves comparable solution quality to the traditional approach but incurs a higher network load. The second design, which completely distributes the population between the workers, reduces network overhead but exhibits lower solution diversity while keeping enough good results in terms of optimization objective minimization. Finally, the proposal with a distributed population and that only interchanges solution between the workers' neighbors achieves the lowest network load but with compromised solution quality.

Figures (8)

• Figure 1: Research domains of the related work and the research gap covered with our proposal.
• Figure 2: Message diagram for the communication in the Semi-distributed GA proposal. The optimization tasks executed by workers and the coordinator are labeled in blue color. The messages are labelled in black color with the format "topic {payload}".
• Figure 3: Message diagram for the communication in the Fully-distributed and Neighbor-aware GA proposals. The optimization tasks executed by workers are labeled in blue color. The messages are labeled in black color with the format "topic {payload}". The optional designs for the initialization and finish phases are blurred.
• Figure 4: Objective spaces of the four experiment scenarios.
• Figure 5: Quantitative evaluation of the minimization of the objectives of the Pareto fronts through the Generational Distance metric.