Can anyone explain what a fused track is from the perspective of radar?
A fused radar track is a track (series of radar detections belonging to the same aircraft) that contains information from detections by multiple radars.
Creating a fused track seems a simple thing to accomplish, however it is not as easy as it may seem.
First of all, lets look at a mono-sensor track. That is a series of detections by a single sensor belonging to the same aircraft. The tracking system has to determine for each detection to which track it belongs to associate the flight data (e.g track label) for that track to the plot (display of the detection on the screen). For primary radar systems (and to lesser extent Mode A/C radar) this can already be difficult when tracks are crossing each other at shallow angles. Especially when you know that the measurements are a bit noisy; an aircraft flying a straight line does typically create a somewhat randomly jagged line.
Now suppose we want to add a second radar to the mix, because the coverage area of a single radar is not big enough. Now we have a situation where there is a common coverage area where the two radars provide overlapping coverage.
In an ideal world, this would show the detections exactly in line with each other. However, each radar has their biases are other measurements error (with various sources, partly external to the radar system). So instead of one track for one aircraft, we end up having two tracks for one aircraft, sometimes a mile apart. This is of course not ideal and can even lead to safety risks.
A simple solution is to have a mosaic system; in such a system there is for each geographical area exactly one radar appointed to provide the surveillance data to the air traffic control's display. The down side of such a solution is that at the edge of the mosaic tiles, when the aircraft moves from one radar area into another, the track may jump because of the differences in measurement errors in both radars. Again not an ideal situation.
That is where radar fusion comes in. In a multi radar tracker, data from multiple radar is combined into a single track. In the process, biases and other measurements errors are greatly reduced, resulting in a track that generally much smoother than a single sensor track. Because of the identification of radar biases, there is no discontinuity (jump) when track move out of coverage of one radar. And due to the higher rate of updates from multiple radars, the update on the radar screen can often be increased as well.
The fused radar track is no longer a depiction of detected positions, but calculated estimate of where the aircraft is based on data from all radar (or other sensors) used in the fusion process. The mathematical process behind the fusion often uses on a form of Kalman Filtering, highly adapted for this task. Outliers are discarded to improve the smoothness of the track.
Trackers like ARTAS or Phoenix do not only fuse radar data into the track, but also include data from ADS-B and Wide-Area Multilateration (WAM). We then speak about a (fused) multi-sensor track instead.
Relative to commercial flight tracking, what this means is that observations from different air surveillance zones are combined to provide a single track of motion for the flight. For example, the major mode S sensing radars in the United States have the following sites:
As you can see there are many different radars. To track a flight that moves across the country, the information from many different radars has to be combined and correlated. This is called "data fusion".