Genetic Algorithm-based Mapper to Support Multiple Concurrent Users on Wireless Testbeds

TL;DR

This paper presents a genetic algorithm-based mapping method to optimize concurrent user access on wireless testbeds, significantly improving utilization and request handling compared to traditional approaches.

Contribution

It introduces a novel genetic algorithm-based mapper that employs spectrum slicing and sub-graph isomorphism to maximize testbed utilization for multiple concurrent experiments.

Findings

Achieves 82.96% of optimal request serving in tests

Serves five times more requests than simple allocation policies

Enhances testbed utilization through topology design and spectrum slicing

Abstract

Communication and networking research introduces new protocols and standards with an increasing number of researchers relying on real experiments rather than simulations to evaluate the performance of their new protocols. A number of testbeds are currently available for this purpose and a growing number of users are requesting access to those testbeds. This motivates the need for better utilization of the testbeds by allowing concurrent experimentations. In this work, we introduce a novel mapping algorithm that aims to maximize wireless testbed utilization using frequency slicing of the spectrum resources. The mapper employs genetic algorithm to find the best combination of requests that can be served concurrently, after getting all possible mappings of each request via an induced sub-graph isomorphism stage. The proposed mapper is tested on grid testbeds and randomly generated topologies. The solution of our mapper is compared to the optimal one, obtained through a brute-force search, and was able to serve the same number of requests in 82.96% of testing scenarios. Furthermore, we show the effect of the careful design of testbed topology on enhancing the testbed utilization by applying our mapper on a carefully positioned 8-nodes testbed. In addition, our proposed approach for testbed slicing and requests mapping has shown an improved performance in terms of total served requests, about five folds, compared to the simple allocation policy with no slicing.