AX25 and TAPR speak might not be as bad as IrDA, but I wouldn't place a bet on that.

I've done plenty of radio modem applications on simple micros. We just wrote our own packet protocol which is cheaper than using a outweight standard you don't require. If you like overweight standards, IrDA will be your Mt. Everest.

The only time I used error correcting codes was in minicomputer disk controller designs. I wouldn't rush in an assume its use until checking out the problem. You can waste a lot of time doing beautiful mathematics when you don't need to do so. But who doesn't waste a lot of project time, really?

(1) Are use using transceivers?

(2) What is the nature of the communications disruption?

One system I did in the 80s required use to send packets through a medium that guarenteed each packet would get hit at least 3 times (10% transmitter outage in acceptance testing: 1 ms of every 10ms the transmitter was keyed off and we had 30ms packets).

We assumed this was to make our company "no-bid" the contract since the requirement was added AFTER we beat two other companies in the preliminary system testings.

We threw together a new packet protocol in 5 weeks with several resync frames inbedded and I wrote a binary tree crcc partial computation routine that took three packets (all guarenteed to have some garbled frames) and its ran every frame permutation between three garbled packets until it got a solution where each frame that passed CRCC had two incidences matching (we voted each frame in).

The routine was so efficient, we could run all permutations even though we prune worthless branches early.

The nature of the 16bit crcc was such that I could compute the entire message CRCC on a Z-80 and prune trees of no possible continuance. When a permutation was rejected, it could stack back out and pick up the next permutation with the partial result on the stack.

We won the contract - we kicked Motorola's butt. :)

-duh