Basic AIS data cleaning

[hughfe]

I'm writing an AIS forwarding program (reinventing a wheel) and I am wondering what is the best practice in cleaning data without losing anything useful. Steps I have done so far are dropping:
less than 21 bytes
checksum missing
checksum invalid
not 7 comma separated words (testing for 6 commas)
if it's a part 1 message, the AIS payload must be at least 7 characters.

[jvde-github]

I would check the length of the message vs message type (1..28). Some serial devices seem to drop a few bits now and then. Then the next step is logical checks, e.g. realistic speed etc. Not sure I would go there for a forwarder.

[hughfe]

Thanks Jasper, the length vs message type is a good idea. As you say sanitising data is a never ending piece of string. I just saw a SART aircraft doing 2 kts and fishing. I think I'll just pass those on and let someone else decide what is worth keeping.