TMA-DPSO: Towards Efficient Multi-Task Allocation With Time Constraints for Next Generation Multiple Access

Abstract

Future heterogeneous services and applications require the provisioning of unprecedented massive user access, heterogeneous data traffic, high bandwidth efficiency, and low latency services in next generation multiple access. In response to the requests from these services and applications, a large number of workers with scattered computing power need to be managed uniformly and scheduled in an efficient manner to perform various tasks. Therefore, task allocation has become a crucial issue to determining whether next generation multiple access can support future heterogeneous services and applications. In this paper, we propose a novel Two-stage Multi-task Allocation method based on Discrete Particle Swarm Optimization (TMA-DPSO). TMA-DPSO is easy to implement and has good search efficiency, which is suitable for large-scale task allocation in next generation networks. Under TMA-DPSO, we redefine the particles in discrete coding form, iteratively update the position and velocity based on the individual optimal particles and the global optimal particle, and finally obtain a corrected optimal solution. Unlike previous methods that only focused on the first-stage task allocation, we make full use of workers’ remaining time to perform second-stage redundant task allocation, which can not only increase workers’ income, but also potentially improve fault tolerance and security. As far as we know, this is the first attempt to utilize the remaining time after first-stage allocation. Finally, we evaluate TMA-DPSO extensively using the synthetic and real-life datasets. The results demonstrate that whether in a compactly or uniformly distributed scene, TMA-DPSO outperforms three benchmark methods by increasing 2.15%-42.24% platform revenue and 6.1%-46.63% workers income.