A new ability to maintain updated data could aid robots in conducting inspections of structures or searching for survivors in disaster-stricken areas.
WiFi is a low-cost and dependable option for deploying time-sensitive applications, such as search and rescue missions, smart factories, and smart city intersections, which increasingly rely on collaborative multi-agent systems. Traditional communication networks are unsuitable for large-scale multi-agent systems that require real-time information sharing to interact efficiently.
MIT engineers have devised a new approach, called WiSwarm, that customizes wireless networks to manage a high volume of time-sensitive data from multiple sources. This prevents data congestion, which could impede real-time reporting essential for robots inspecting buildings or searching for survivors in a disaster zone. WiSwarm facilitates cost-effective communication among numerous robots over WiFi networks without requiring costly communication and processing hardware.
The MIT team, including authors Vishrant Tripathi, Ezra Tal, Muhammad Shahir Rahman, Alexander Warren, Sertac Karaman, and Eytan Modiano from the Laboratory for Information and Decision Systems (LIDS), and Igor Kadota SM ’16, PhD ’20 from Columbia University, will present their novel approach at IEEE’s International Conference on Computer Communications (INFOCOM) in May. The approach enables multiple robots to communicate via existing WiFi networks, eliminating the need for expensive and heavy onboard communication and processing hardware.
The team’s method differs from conventional networking protocols prioritizing first-come-first-serve data processing. Instead, the approach prioritizes the most recent data for time-sensitive tasks like detecting moving objects. According to Tripathi, “all the old video frames are useless. What you want is the newest video frame.”
To ensure data freshness, the MIT team proposed prioritizing the “age-of-information” metric, which measures how up-to-date the data is based on the application’s needs. The team aimed to incorporate a “last in, first out” protocol for multiple robots working on time-sensitive tasks via conventional wireless networks, which are more accessible and don’t require bulky communication hardware.
However, wireless networks have a distributed nature that can lead to data congestion when multiple sources send data simultaneously. Even with a “last in, first out” protocol, data collisions can occur, leading to a breakdown in time-sensitive applications.
The team utilized their algorithm with mobility-tracking drones, equipping them with a camera and Wi-Fi-enabled chip to relay images to a central computer. Pairing the network with the algorithm allowed the computer to receive the latest images and send commands to keep the drones on track. Testing the system with two drones resulted in twice as fresh data, leading to six times better tracking. Wi-Fi alone couldn’t handle the data traffic when expanded to five drones and vehicles, causing the drones to lose track. WiSwarm improved the network and allowed all drones to track their vehicles.
The MIT team developed WiSwarm, a scheduling algorithm that manages multiple data streams over wireless networks. The algorithm determines which source to send data next by considering three parameters: priority, information age, and channel reliability. By scheduling drones to report updates one at a time, the system provides the freshest data for time-sensitive tasks without clogging the network. In tests, WiSwarm relayed data that was two times fresher and enabled six times better tracking than using Wi-Fi alone with two drones. With five drones and five ground vehicles, WiSwarm was better equipped and enabled all drones to keep tracking their respective vehicles.