After we recorded real data for the transmitter operation over the last week, I used the existing Simulink model of the receiver to understand the challenges posed by such a strongly fading environment to the system performance. It was easy to see that the very large coherence time of the channel (on the order of 100 ms), as well as the relatively large amplitude of the delayed echoes, drives the error probability at the receiver to very high values, even though the symbol synchronisation algorithm stabilises quickly. While this is likely less of a problem in a real shallow water channel, due to the absence of back- and side-reflecting surfaces, it still encourages us to implement a frequency hopping and coded modulation scheme. This is what we will be investigating over the next week, while implementing the more generic blocks of the receiver.

We also found out there was a firmware bug in the hydrophones we used for recording (the digitalHyd SR-1 from MarSensing), which led to some artifacts in the data. We will have to perform additional tests to understand how much of an impact this had on the simulated receiver performance.

There was no more work done this week, as I took the remainder of the days off as an Easter vacation. Back to regularly scheduled work on Monday.