Dynamic load balancing on heterogeneous clusters for parallel ant colony optimization

Abstract

Ant Colony Optimisation (ACO) is a nature-inspired, population-based metaheuristic that has been used to solve a wide variety of computationally hard problems. In order to take full advantage of the inherently stochastic and distributed nature of the method, we describe a parallelization strategy that leverages these features on heterogeneous and large-scale, massively-parallel hardware systems. Our approach balances workload effectively, by dynamically assigning jobs to heterogeneous resources which then run ACO implementations using different search strategies. Our experimental results confirm that we can obtain significant improvements in terms of both solution quality and energy expenditure, thus opening up new possibilities for the development of metaheuristic-based solutions to ``real world" problems on high-performance, energyefficient contemporary heterogeneous computing platforms.

SWITCH workbench: A novel approach for the development and deployment of time-critical microservice-based cloud-native applications

“…Most adaptation research has focused on finding solutions for systems that use heterogeneous infrastructure but homogeneous components. For instance, A. Llanes et al [27] developed a system to balance ant colony optimisation tasks on heterogeneous infrastructure. P. Jamshidi et al [28] presented a system based on fuzzy logic and the vPerfGuard [29] team developed a system that can predict performance based on low-level metrics.…”

Re-engineering the ant colony optimization for CMP architectures

“…Concerning distributed ACO, Stützle [26] introduced the first and most simple strategy, in which parallel independent runs of the algorithm were executed, and the best solution of the runs was taken as the final solution. This parallelization scheme, with no communication overhead, was followed in [27,6] targeting GPU-based clusters. Approaches based on the island-model, in which the nodes (colonies) exchange information after every certain number of iterations, have also been proposed [28,29].…”

“…Therefore, the research community has provided several ways to optimize the ACO algorithm for TSP (ACO-TSP) in High Performance Computing (HPC) architectures. The first attempts were carried out on NVIDIA GPUs using CUDA [4][5][6][7], and more recently on the first generation of Intel Xeon Phi (Knights Corner) [8][9][10]. Intel architectures offer several advantages compared to Nvidia architectures.…”

Peer Review #1 of "Enhancing the performance of the aggregated bit vector algorithm in network packet classification using GPU (v0.1)"

“…Several investigations including (Li et al 2013;Lin et al 2016) have attempted to study the use of CUDA platform for parallel implementation of network functions such as IP lookup in routing tables, aiming at having access to higher throughput. This tool has been used for executing parallel genetic algorithms (Zhao et al 2018), neural networks (Gong et al 2017) and ant colony optimization algorithm (Llanes et al 2016). Also, the capacity of parallel PeerJ Comput.…”