Improving Age-of-Information in Distributed Vehicle Tracking

We consider the problem of tracking vehicles in a distributed setting, which is important for, e.g., autonomous driving and collision avoidance systems. These applications rely on receiving timely updates and require aggregating sensor data from many sources to improve accuracy. We consider a cloud-assisted scheme that utilizes replication to alleviate the straggler problem, i.e., the problem of random delays in distributed systems. We derive the age-of-information (AoI) of estimate updates and show that replication significantly improves the AoI. Furthermore, we derive the probability that the error of the position estimate exceeds some threshold for a given AoI.