Adaptive scheduling for multi-objective resource allocation through multi-criteria decision-making and deep Q-network in wireless body area networks

Document Type


Source of Publication

Journal of Ambient Intelligence and Humanized Computing

Publication Date



To provide compelling trade-offs among conflicting optimization criteria, various scheduling techniques employing multi-objective optimization (MOO) algorithms have been proposed in wireless body area networks (WBANs). However, existing MOO algorithms have difficulty solving diverse multi-objective optimization problems (MOPs) in dynamic and heterogeneous WBANs because they require a prior preference of the decision makers or they are unable to solve non-discrete optimization problems, such as time slot scheduling. To overcome this limitation, in this paper, we propose a new adaptive scheduling algorithm that complements existing MOO algorithms. The proposed algorithm consists of two parts: scheduling order optimization and the auto-scaling of relative importance. With the former, we logically integrate the decision criteria using a multi-criteria decision-making (MCDM) method and then optimize the scheduling order. For the latter, we adaptively adjust the scales of the relative importance among the decision criteria based on the network conditions using a deep Q-network (DQN). By tightly integrating these two mechanisms, we can eliminate the intervention of decision makers and optimize non-discrete tasks simultaneously. The simulation results prove that the proposed scheme can provide a flexible trade-off among conflicting optimization criteria, that is, a differentiated QoS, reliability, and energy efficiency/balance compared with a conventional approach.




Springer Science and Business Media LLC


Computer Sciences


Adaptive timeslot scheduling, Auto-scaling mechanism, Deep Q-network (DQN), Multi-criteria decision-making (MCDM), Multi-objective optimization (MOO), Wireless body area networks (WBANs)

Indexed in Scopus


Open Access