The need for a high-bandwidth communication link to carry multiple video channels from a mobile robot (unmanned ground vehicle, or UGV) back to a control station requires using high-frequency RF communication links. These links are mostly line-of-sight, often limiting the flexibility of the robot’s movement. The Autonomous Mobile Communication Relays project demonstrated the capability of autonomous slave robots to provide communication-relaying capability for a lead robot exploring an unknown environment.
Several generations of Automatically Deployed Communication Relay (ADCR) systems led to a robot-mountable Deployer module that deploys static relay nodes when and where needed. To meet specific in-theater requirements, the Manually Deployed Communication Relay (MDCR) system was developed. A fourth-generation ADCR system was developed in 2013 to automate the deployment of these fielded and proven MDCR relay nodes.
Radio Frequency Principles
One of the common weaknesses in other relay systems that causes them to fail in field tests is the use of high-gain antennas. While high-gain antennas may seem desirable for range extension, without the use of an electronic amplifier, this gain can only be achieved by focusing the radiation pattern.
For a relay system to be versatile and work well in various environments, low-gain antennas are generally best. Low-gain omnidirectional antennas should be used for a relay system to be versatile and able to work in a wide variety of environments with no a-priori terrain and placement information.
The height of the relay-node antenna when placed on the ground must be equal to or greater than the height of the antenna on the robot. Otherwise, the deployed relay node with a lower antenna would encounter lower received signal strength than the robot, and might be unable to join the network. When two nodes are in close proximity, the receiver front end of one node tends to get saturated by the strong signal emitted by the nearby node’s transmitter. This may lead to mutual jamming so that neither can enter the network. This often means that only one node should be on at a time while being transported by the robot.
MDCR relays and end-point radios use standard half-wave dipoles with 2.1- dBi gain. Active MDCR nodes (operating at 4.9 GHz) must be kept at least approximately 1 m (40”) from each other to ensure no mutual jamming. For this reason, in the first- and second-generation ADCR systems, only one stowed node inside the Deployer module was active at any given time. The system ensured that the active node had successfully joined the network before deploying it and activating the next node in the Deployer module. In the MDCR design, both relay nodes had to be active while being carried by the robot. Placing the two nodes on a level tabletop at the same distance apart prevented them from entering the network.
Input filtering is the best defense against unintentional radio frequency (RF) jamming to preserve the link between the operator control unit (OCU) and the robot. A commercial bandpass filter was used on the OCU-side MDCR end-point radio to mitigate jamming issues. The center frequency and bandwidth of the bandpass filter were chosen to match the relay network’s frequency characteristics. The bandpass filter helps to attenuate the noise outside the frequency band of interest, improving the overall signal-to-noise ratio of the received signal.
The high throughput required to transfer multiple streams of video data from a remotely controlled vehicle (often outfitted with multiple cameras) limits the practical number of relay nodes in the data route. This limit is about three for 802.11g wireless networks. Techniques for increasing this limit include reducing the video resolution, eliminating the color component, or using multi-frequency radios.
Due to the delay incurred in establishing a new route, the constant switching between two routes could bring the network to a halt. One way to prevent this is to use some measure of hysteresis and “good enough” metrics so that a new route is not selected as long as the current route can still carry the required network traffic.
The required throughput, type of data, mesh topology, and the data usage determine the maximum number of relay nodes that can be used in a data-traffic route. Problems with remotely controlling the vehicle in real time begin to appear after three relays are present in a linear route. This can be mitigated by reducing video resolution, number of cameras, and/or dropping color information from the video stream. Another solution is to use dual-frequency radios to allow simultaneous transmission and reception of data at each node, increasing the overall throughput capacity of the mesh network.
Several key attributes are desired in a relay-node antenna mast deployed from a UGV. Appropriate antenna height is considered critical. Additionally, resistance to impact loads and durability are also very important. Deployment of the relay node may be while the UGV is in motion and may include a free-fall to the ground. For this reason, the relay node and antenna must be constructed to mitigate the destructive effects of ground impact. An antenna mast can be constructed to withstand the impacts or stowed within an outer shell that protects it until after the node has been placed. The antenna can then be deployed.
Deployable antenna masts are usually more complex because they need to transform from a compact shape inside the relay node enclosure into a long, straight, vertically oriented mast. Many different concepts of deployable antenna were examined to understand their utility. Nine types of antenna masts were analyzed, and of these designs, the spring-hinge and spring-steel foldable masts are most suitable for relay nodes to be deployed from moving unmanned ground vehicles.
A spring-hinge antenna mast is made of three aluminum links, a radiating element, and four spring hinges. The mast is folded and stored in a cavity in the relay node for protection until the relay node has come to rest on the ground surface. This type of antenna mast proved effective for deployment from vehicles the size of an iBot® Packbot® UGV. The antenna mast was approximately 18” tall and could be folded into a relay node less than 8" long. This design could be scaled to larger systems. The relay enclosure protected the antenna effectively before mast deployment. However, the complexity of the mast reduced its durability.
A spring-loaded linear telescoping mast with stacked links was developed as a conceptual prototype. It was very compact, but also very complex and prone to jamming. A spring-steel foldable antenna was used with the MDCR system. The antenna mast is rugged and flexible; it can be bent 180° at one point and still return to its original position. It is constructed from two long strips of spring steel, each with a parenthesis-shaped cross section (similar to a tape measure). The two pieces of spring steel are held together with an outer sheath, and the coaxial cable for the antenna runs between the pieces of spring steel. This type of antenna mast, designed for use with UGVs, is rugged, reliable, and can be configured to reach a beneficial height.
Relay Node Deployment
A relay-node enclosure contains the radio, battery, and electronics, and provides for antenna deployment. It must also be designed to meet the shock, vibration, ultraviolet, thermal, and ingress protection requirements of the application. Interface requirements may also need to be considered, such as recharging, power switch, battery level, relay retrieval, remote power toggling, and channel selection.
The robot carries a Deployer module that contains the relay nodes to be deployed. Since a relay node takes a finite amount of time to fully boot and enter the network, it is of utmost importance to have a relay node that is fully booted and in the network before deployment is needed. This prevents any network interruptions because of deployment.
All deployment systems require a method for placing the relay node on the ground. The first-generation ADCR system (Figure 1) used a compression spring-loaded mechanism. The second-generation ADCR system (Figure 2) used a constant-force spring-loaded mechanism. MDCR used a fork with magnets to hold the relay node securely, with the node placement accomplished by the robot flippers. Finally, the fourth-generation ADCR (Figure 3) used a motorized forked carrier with magnets to place the relays. Many other methods are conceivable, such as dropping the relays out of the bottom of a Deployer using gravity. The spring-loaded designs had the benefit of allowing relays to be closely packed into a small space on top of the robot. Shock isolation is an important consideration when designing a relay node that will be dropped from a fast-moving UGV or from a considerable height. For this reason, internal electronics may need to be mounted on shock and vibration isolators.
An Ethernet link is normally used to communicate between the robot and the relay-deployment module it carries; however, Ethernet is not a strict requirement. If the robot is one of the older analog systems, then a video/audio codec board can be used to convert the analog signals to Internet Protocol (IP) Ethernet data.
One of the common issues with initial prototype systems was that the robot and OCU needed to be reconfigured for every test. This was an unfriendly, unreliable, and time-consuming process. Additionally, this is unacceptable for a system to be fielded. The systems had to be plug-and-playable, with no configuration needed on the target robot or OCU. This started a search of various technologies to make connecting the robot and OCU over a mesh simpler, and resulted in the use of VPN technology, specifically, OpenVPN. This technology provides a wrapper around the network messages, providing a plug-and-play solution. The radios do not need to know which robot or OCU generates or receives the traffic data; hence, no robot or OCU configuration is needed. The drawback, however, is that data generated by one radio can only be received by the other in the pair, while the intermediate relay nodes can be any compatible mesh node.
This article was written by Hoa Nguyen, Narek Pezeshkian, Aaron Burmeister, and Abraham Hart of Space and Naval Warfare Systems Center Pacific (SSC Pacific), San Diego, CA. For more information, Click Here .