ETA Calculation Simulations

Using data obtained from Google Maps, we were able to simulate how long it took to drive from one checkpoint to another with and without traffic. The following tests were designed to determine the accuracy of the GPS algorithm being utilized by the STaR System.

We knew how long it would take to get from a checkpoint to a shuttle stop, so using this data, we were able to calculate the difference against the calculated values. The graphs below show the errors in calculation as a shuttle progressed through the various checkpoints.

Simulation Set 1

Average Error: 1.008889 minutes

This first set of 10 simulations simulated the typical early morning behavior of a shuttle when there was a slight traffic for several trips. Also notice that even though the errors were higher at checkpoint B, it gradually lowered as more checkpoints were passed.

Simulation Set 2

Average Error: 1.515556 minutes

This set of data was slightly more unrealistic that the previous set because instead of traffic building up (or building down), travel time just abruptly changed. Although the system eventually compensated for the errors, it took a little longer to do so. Simulations 8 to 10 also simulated transmission errors where some of the checkpoints were not transmitted at all.

Simulation Set 3

(Note: Simulations 1 to 5 had GPS errors, so one or more packets were transmitted more than once. Simulations 6 to 10 had transmission errors.)

Average Error (GPS Errors): 1.562222 minutes

Average Error (Transmission Errors): 1.536667 minutes

There might have been a higher average error when the system encountered technical problems, but just like in the previous tests, as more checkpoints were transmitted, the accuracy increased.

Even when there were a significant amount of transmission errors and only a couple of checkpoints were transmitted, the error was still acceptable. These last set of simulations were not very likely to happen, but even if they did, the system still handled the ETA calculations relatively well.

Radio Modem Signal Strength Along the Route

In order to determine the best placement for the receiver module on campus, two tests were conducted. In the first test, the receiver was placed right by the campus shuttle stop, and then at the very top of the Cesar Chavez Student Center building. The shuttle stop was chosen for the test because it would be a convenient location to install the receiver module as it minimizes further data transfer of the ETA calculation. The student center building was chosen for testing because it was the tallest structure on campus that was easily accessible.

The test was conducted with one of the radio modems constantly broadcasting a test signal. The other radio modem was constantly receiving the test signal, and the signal strength of the test signal was noted at the shuttle’s checkpoints. The graph below shows the results of the test.

Even though there was stronger reception on average by the campus shuttle stop, it was still not the ideal location because the radio modems completely lost contact once a shuttle got past John Daly Blvd. Signal reception stopped past John Daly Blvd because the receiver on campus was obstructed by the Park Merced buildings and the concrete parking structure at the Daly City Bart station.

When the receiver was on top of the student center, the signal strength was overall not as strong, but there was reception, albeit weak, in all of the checkpoints. There was signal strength throughout the route because the receiver was in a slightly higher elevation and the signal was not obstructed by all the buildings in Park Merced. However, the signal strength was not as strong because the library provided some obstruction to the signal.

The tests verified that the radio modems were strong enough to cover the entire route of the shuttle. However, the radio modems needed to be in a tall building to cover the entire route. If the radio modems were also in a tall building that was also not obstructed by the other buildings along the route, then a good data network can easily be established to increase the effectiveness of the whole system.

Another test outside the route was also conducted just to see if there were any potentially good areas to place a receiver module on the Daly City Bart station. With the a radio modem on top of the student center and another on top of the Bart parking garage, a signal strength of -93dB was measured which was relatively good considering that it was almost the same value as when the transmitter was closer to campus. From this result, we concluded that if only the receiver module was placed in an even higher structure on campus such as the Admin building or at any one of the Park Merced buildings, then there would have been easily a consistently strong signal throughout the shuttle’s route.

Comments are closed.