Local State Scoring Relative to Global Goals for Self-Organising Unmanned Systems in Contested RF Environments
Funding: DST Group Research Agreement
RF signals can be degraded or lost by environmental effects such as shadowing, fading and interference or by an adversary who actively seek to disrupt communications. In such environments, additional data ferrying UAVs used in conjunction with delay tolerant networking protocols can be used to build communication capabilities. The communications-management overhead between these nodes needs to be small to avoid excessive use of the already limited bandwidth. Research is underway into self-organisation techniques with autonomous UAVs using simple local rule-sets to form emergent, complex and global behaviours. This avoids the need to centrally coordinate data ferrying nodes.
In order for a given node to have an understanding of how well it is performing in its attempt to achieve a global goal, it must be able to score how well the rule-set it is implementing is performing within the current environmental context which can be observed locally. That is, based on local observations a node must decide whether or not it is performing well on meeting a global outcome. The local node is not in possession of a global view, nor will it solicit such information from the network (the solution must be scalable, and not place undue burden on the limited resources). Such scoring of the rule-set in a given context (the environment) for a given mission is essential to enable reasoning and decision making as to “which rule-set should a node implement given its mission and the state of its observable environment.”