Packetizing voice for mobile radio

In present mobile radio systems, conventional FM techniques are used to transmit speech in its analog form. A number of digital schemes have been proposed in the past as an alternative. To overcome the effects of Rayleigh fades that appear in the received speech as noise bursts in the form of “pops” and “clicks”, the author proposes packetizing speech and encoding the packets in an error-detecting code before transmitting them. If the receiver detects an error, it throws away the packet and requests the transmitter to retransmit the same packet. If the requested packet has not arrived in a given time period, the missed packets are replaced with zero amplitude values. The variable delays that packets are subjected to as a result of this procedure are smoothed out before packets are played back so that they all appear contiguous at the receiving end. The resulting improvement in the SNR at the receiving end depends, among other things, on the maximum delay and the vehicle speed, and is achieved in addition to any improvement that is possible with a specially designed coder. For example, with 4 ms long packets and 12 ms of delay, the SNR improves by about 13 dB at a vehicle speed of 35 mph and 17 dB at 12 mph