Swarm Based Implementation of a Virtual Distributed Database System in a Sensor Network

Abstract

The deployment of unmanned aerial vehicles (UAVs) in recent military operations has had success in carrying out surveillance and combat missions in sensitive areas. An area of intense research on UAVs has been on controlling a group of small-sized UAVs to carry out reconnaissance missions normally undertaken by large UAVs such as Predator or Global Hawk. A control strategy for coordinating the UAV movements of such a group of UAVs adopts the bio-inspired swarm model to produce autonomous group behavior. This research proposes establishing a distributed database system on a group of swarming UAVs, providing for data storage during a reconnaissance mission. A distributed database system model is simulated treating each UAV as a distributed database site connected by a wireless network. In this model, each UAV carries a sensor and communicates to a command center when queried. Drawing equivalence to a sensor network, the network of UAVs poses as a dynamic ad-hoc sensor network. The distributed database system based on a swarm of UAVs is tested against a set of reconnaissance test suites with respect to evaluating system performance. The design of experiments focuses on the effects of varying the query input and types of swarming UAVs on overall system performance. The results show that the topology of the UAVs has a distinct impact on the output of the sensor database. The experiments measuring system delays also confirm the expectation that in a distributed system, inter-node communication costs outweigh processing costs.