User login

Weekly Report

31

May

2013

The PHY and channel objects are pretty much done and for the last little while I have been throughput testing and debugging. Initially using a 1.155 Gbps PHY I was reaching only 33 Mbps of measured throughput with a 1500 octet TCP bulk sending application. Moving to UDP speed things up to 100 or so Mbps. Converting the transmit duration calculations from microseconds to nanoseconds speed things up again. After that it was a lot of head banging to realise that the single line of throughput reporting was outputting in Mebibits instead of Megabits. Also getting the exact time in nanoseconds of the UDP flow from when the first packet is received to the last helped bring in the reported throughput to less than 1% difference from the theoretical throughput based on the simulations average number of backoff slots. This difference should be accounted for due to beacon transmissions.