A dendritic cell-mediated framework for optimal resources scheduling and planning

Abstract

This research presents the development of a novel computing paradigm inspired by the remarkable features of the human immune system with the capability of solving real-world scheduling and optimization problems. Specifically, the development of the dendritic cell (DC) inspired framework, namely, the DC-mediated Signal Cascading Optimization Framework, is described. The DC inspired framework is an emerging computation paradigm in Artificial Immune Systems (AIS) based on the functions and behaviours of the dendritic cells in the human immune system. Biologically, dendritic cells are vital in performing sentinels, regulatory and decision-making operations in optimising the T-cell priming for the adaptive immunity in a complex metaphor of maturation and migration. By focusing on the decision making mechanism in DCs, effective decision can be regulated by the processes of interpretation of environmental instructions and intrinsic programs, which decomposes the steps of (i) the integration and quantification of a multitude threat signal/stress, (ii) interactions with receptors, (ii) the downstream cellular signal cascading or transcriptional pathways, and (iii) responses control. These aforementioned features and desirable behaviour form part of the DC-mediated framework. As such, the transcriptional pathways and control functions is being formulated to signal-regulated solution evolving mechanisms inherent in the DC-mediated framework. In the context of this research, a DC-mediated framework is developed and applied to solve a real-life resource scheduling and optimization problem, which is an NP-hard problem. The case study represents a typical resource management problem in an air cargo terminal where the operations of air cargo occur in a highly complex and dynamic manner that is analogous to that of the human immune system. The problem is modelled and implemented in the framework, and tested with real operation data. Near-optimal or optimal solutions are generated by the framework and the results are evaluated with regards to enhance the efficiency of using the resources. According to extensive experimental studies, including sensitivity analysis and benchmarking studies, the research reveals the capability of the framework in pursuing high quality solutions, in particular, for large-scale problems. In addition, the significance and the roles of the intrinsic effector control functions in the framework are identified.

As such, the research presented in this thesis not only demonstrated the problem solving capability and scalability of the DC-mediated framework as an optimization tool, when comparing the results with solutions adopted in real operations, but also extended the knowledge in the domains of artificial intelligence and logistics engineering.