Sherrill E. Todd. Qualifying situation awareness in Janus

Military Academy West Point, NY, 1998. — 53 p.We have developed a measure of the theoretical increase in situation awareness of a tactical commander as a result of receiving data from reconnaissance, scouting and intelligence activities. We intend the measure, called information gain (IG), be used as a measure of effectiveness of information systems. It is based on the concept of modeling a commander's uncertainty about his adversary's disposition in terms of probability distributions over the set of states the adversary may occupy. Starting with an initial distribution, subsequent updates are calculated using Baye's formula, exploiting the operating characteristics of the sensor systems used and the search activities conducted during a sequence of time intervals. In particular, the updating process requires the probability of detection and probability of false alarm for each set of parameters involved, including the time interval, the sensors used, the area scanned by the sensors and the targets possessed by the adversary and their positions. We implemented computation of information gain for combat simulations conducted using the Janus model. This proved to be a challenge for Janus simulated combat because we had to devise innovative methods to obtain detection probabilities for sensor target pairs at various ranges, and also to determine the sets of cells within the battle area scanned by each sensor during each time interval. We also developed ways to account for movement of mobile targets over time. For example, a target located at a certain time may not be in the indicated location at some later time, provided the target is not killed and it is not re-detected. Thus, information about such a target may actually degrade over time. In this report we describe our approaches to these and other issues in implementing the information gain measure for JANUS applications.