Conventionally, most autonomous mobile nodes, including unmanned aerial vehicles (UAVs), are operated remotely by human operators. As such, forming a mobile communication network to connect several airborne or ground-based nodes with multiple remotely controlled UAVs can be prohibitively costly and very inefficient because many operators are needed. Furthermore, the operators need to control the deployed UAVs, while coordinating with others and monitoring the performance of the deployed communication network. Hence, a feasible solution must support collaborative communication, sensing and navigation. BioAIR offers a solution to this problem by enabling multiple UAVs to autonomously control themselves based on network performance.
This technology can assist with a variety of commercial and military applications. For example, collaborative communication is required for applications such as communication in area-denial environments and high-bandwidth Internet in the sky. Similarly, collaborative sensing is required for applications such as communication in anti-access environments, remote sensing, geolocation, aerial surveillance, and border patrol. Collaborative navigation is required for applications such as target tracking, agricultural product care, mobile escorting, and perimeter defense.
A successful solution to this problem must address several challenges. The first challenge is the limited range of wireless communications. To overcome this challenge, nodes must route their data traffic through their neighbors. However, in contrast to standard wireless networks, this is a mobile ad-hoc network with highly volatile connections due to environmental interference and frequent connects/disconnects. The BioAIR algorithm uses the strength of the wireless signals received by a node to determine its flying behavior, thereby allowing a node to control its connections/disconnections and minimize the chances of environmental interference intelligently.
The second challenge is coordinating the motion of nodes in a distributed manner and preventing the formation of suboptimal networks. BioAIR converts received wireless signal qualities from adjacent nodes into attractive and repulsive fields based on certain thresholds. By summing the strength and direction of these fields with virtual attraction fields from certain types of nodes that are out of range, BioAIR is able to control each node independently to accomplish the overall mission in a distributed manner. The third challenge is flying heterogeneous airborne nodes from one point to another in a safe manner.
Since most of UAVs have some limited autopilot capabilities based on GPS guidance, BioAIR commands the autopilot by monitoring the proximity sensors or the signal strengths from neighboring nodes to ensure both safety and success of the mission.
The fourth challenge is enabling one or more operators to monitor multiple nodes through a unified interface. BioAIR addresses this issue by using the formed ad-hoc network to relay control data back to centralized location and presenting a graphical user-interface to interact with nodes when necessary.
The fifth challenge is seamlessly dealing with node failures. BioAIR utilizes any spare nodes to proactively reinforce critical locations of the network to minimize the repair time in the event of failures.
The sixth challenge is maximizing the effectiveness of the radio network. BioAIR addresses this challenge by adjusting the position of nodes based on the criticality of data traffic.
In addition to these six main challenges, there are several other factors that must be considered in a successful solution. For example, the response time of the algorithm must be a few seconds or less, as the airborne nodes can fly several meters in that time. Similarly, the algorithm must minimize its communications overhead so as to reduce the network load. It must also minimize resource usage due to limitations on the amount of processing, memory, and power onboard a node.
BioAIR assumes the availability of communication data and GPS location information (or a compass for direction), but no additional sensors are assumed. The scope is limited to omnidirectional radio antennas, although unidirectional antennas could be utilized with some additional overhead. This solution can be extended to three dimensions with little effort.
Some key innovations of BioAIR include the mapping of wireless signal strength (e.g. Wi-Fi strength from off-the-shelf hardware, or any similar sensor modality) to field strength, and the differentiation of fields based on two states – namely, the global state of the system and the local state of the node. The global state of the BioAIR system is maintained and broadcasted to every node in the swarm periodically, while the local state is broadcasted to each node’s immediate neighbors. Also BioAIR nodes must have unique IDs or IP addresses for routing communications.
The bio-inspired BioAIR algorithm is designed and developed specifically for controlling a swarm of nodes. BioAIR focuses on forming a wireless communication network between sites of interest, tracking node mobility, and self-repairing in the presence of damaged networks in a distributed manner.
BioAIR was developed to support both fixed-wing and rotary-wing UAVs. However, the majority of testing was performed on a “hexarotor” rotary-wing platform shown in Figure 1. The payload attached to a hexarotor included a minicomputer, a Wi-Fi card connected to it, and a dedicated power pack. The hexarotor’s autopilot runs on a separate minicomputer powered by a different supply. The minicomputers communicate via an Ethernet connection through a software interface that provides GPS coordinates and altitude through an onboard GPS chip, and accepts target GPS waypoint coordinates.
In addition to hexarotors, there were some ground-based destination and origin nodes. In this set of experiments, such a node is represented by a minicomputer or standard PC computer equipped with a GPS chip. Prior to performing live field tests, BioAIR was thoroughly tested in simulation with hardware-in-the-loop.
The base case for forming a communications network is to establish the connectivity between two sites. An origin and a destination are networked together by forming a chain, or a tentacle, that is comprised of multiple airborne nodes. Nodes can be launched from anywhere at varying time intervals with some a priori knowledge about the location of an origin and a destination. Once connectivity is established between a set of nodes, the received signal quality is used to improve the overall performance of the network rather than the straight-line distance between them. It is important to note that typically, the signal quality has exponential falloff with distance, which can change due to environmental conditions.
The idea behind this algorithm is to form tentacles from one or more origins to one or more destinations by growing them in a biological way through the accumulation of nodes. In basic terms, initially all nodes will fly towards an origin. When a node encounters a tentacle or an origin, it will fly towards the destination, and at the edge of its communication range to the existing tentacle or origin, it will hold position if it is rotary-wing, or circular fashion if it is fixed-wing, thus extending the tentacle towards the destination, as shown in Figure 2. When the tentacle reaches the destination, any extra nodes will construct new tentacles or reinforce existing nodes.
The BioAIR algorithm stops a node at the edge of its communications range to a neighbor by converting the received signal quality (SQ) into a field strength, which corresponds to being attracted or repelled by the nearest neighbor. When the nodes are close to each other, they repel to prevent collisions; when they are outside the acceptable range, they attract; and when they are completely out of range, there is no signal or field. At the onset, the destination’s location is disclosed to each node to allow computation of a virtual attraction field even when it is out of range. Obtaining consistent signal quality measures is important to prevent the fields from changing drastically and causing erratic trajectories. Figure 3 depicts real signal quality measures obtained between two airborne nodes during live field tests.
This article was written by Bo Ryu, Nadeesha Ranasinghe, and Wei-Min Shen of EpiSys Science, Poway, CA; and Kurt Turck and Michael Muccio of the Air Force Research Laboratory, Rome, NY. For more information, Click Here