Random linear packet coding for broadcast networks

Random linear packet coding is considered for efficient broadcasting in networks with large propagation delays, such as underwater acoustic networks. To additionally overcome the effects of fading, we combine packet coding with adaptive power control, whereby the transmitter adjusts its power upon receiving feedback from the receiver on the current state of the channel (locally-averaged, large-scale gain). We investigate two power adjustment rules: the worst link rule and the average link rule. In the first case, the transmit power is adjusted in accordance with the link that has the lowest channel gain, while in the second case, the power is adjusted in accordance with the average of the gains on all links. System performance is evaluated based on the average energy per bit of successfully transmitted bit of information, using (i) simulated channels, (ii) experimentally recorded gain values from the MISSION 2012 experiment, and (iii) actual network deployment from the MISSION 2013 experiment conducted off the coast of Singapore. Results indicate energy savings on the order of several dB compared to systems that do not use power control.