Cellular latency with ConnectPort X4

Hi all,

I’ve just upgraded my application from using a ConnectPort X2 and a separate 3G cellular router to using a ConnectPort X4 with an integrated GPRS modem. After several quite odd errors (couldn’t open the same port numbers, quite strange), the Python application is up and running on the X4.

However, when I send a command, what used to happen within seconds now takes almost a minute to occur. It’s quite strange that there’s that much latency. I open a TCP connection to the Python program remotely, send it a command, and don’t see the results for 45 to 90 seconds.

I expected an extreme throughput decrease going from 3G to GPRS, but not quite that bad a hit on the latency.

Any ideas?


This post is on the wrong forum - DigiMesh 900 folk aren’t likely to know or care :slight_smile:

Really this is a complex carrier/tower issue. You may have a tower which doesn’t really support GPRS anymore (AT&T started killing it off a few years ago). Products like the X4+GPRS can’t even be certified anymore, so it exists because it grand-fathered the ban on new GPRS products.

45-90 seconds is pretty bad. My own tests with 2.5G Edge shows 3 to 10 seconds a common range for the initial ‘unpark’ and allocation of band-width. But TCP/Ip requires not only the unpark but the whole 5-way socket-open hand-shake has to succeed. if you can run WireShark on the remote end, it might be interesting to see if the socket open if suffering serious retry issues, adding to your problems.

Can you use UDP/IP? My own tests show it is quite reliable - say 1 or 2 errors per 10,000 packets.