Quick overview of what's here: * gmsk_test.py: stand-alone program that exercises the GMSK packet tx and rx code. The two halves are connected with a simulated noisy channel. It's easy to add extra instrumentation to log various internal states. We used a variant of this code to get this working in the first place. * benchmark_gmsk_tx.py: generates packets of the size you specify and sends them across the air using the USRP. Known to work well using the USRP with the Flex 400 transceiver daughterboard. You can specify the bitrate to use with the -r command line parameter. The default is 500k. Some machines will do 1M or more. * benchmark_gmsk_rx.py: the receiver half of benchmark_gmsk_tx.py. Command line arguments are pretty much the same as tx. Works well with a USRP and Flex 400 transceiver daughterboard. Will also work with TVRX daugherboard, but you'll need to fiddle with the gain. See below. Prints a summary of each packet received and keeps a running total of packets received, and how many of them were error free. There are two levels of error reporting going on. If the access code (PN code) and header of a packet were properly detected, then you'll get an output line. If the CRC32 of the payload was correct you get "ok = True", else "ok = False". The "pktno" is extracted from the received packet. If there are skipped numbers, you're missing some packets. Be sure you've got a suitable antenna connected to the TX/RX port on each board. A "70 cm" / 420 MHz antenna for a ham handi-talkie works great. These are available at ham radio supplies, etc. The boards need to be at least 3m apart. You can also try experimenting with the rx gain (-g command line option). Generally speaking, I start the rx first on one machine, and then fire up the tx on the other machine. The tx also supports a discontinous transmission mode where it sends bursts of 5 packets and then waits 1 second. This is useful for ensuring that all the receiver control loops lock up fast enough. * tunnel.py: This program provides a framework for building your own MACs. It creates a "TAP" interface in the kernel, typically gr0, and sends and receives ethernet frames through it. See /usr/src/linux/Documentation/networking/tuntap.txt and/or Google for "universal tun tap". The Linux 2.6 kernel includes the tun module, you don't have to build it. You may have to "modprobe tun" if it's not loaded by default. If /dev/net/tun doesn't exist, try "modprobe tun". To run this program you'll need to be root or running with the appropriate capability to open the tun interface. You'll need to fire up two copies on different machines. Once each is running you'll need to ifconfig the gr0 interface to set the IP address. This will allow two machines to talk, but anything beyond the two machines depends on your networking setup. Left as an exercise... On machine A: $ su # ./tunnel.py --freq 423.0M --bitrate 500k # # in another window on A, also as root... # ifconfig gr0 10.10.10.1 On machine B: $ su # ./tunnel.py --freq 423.0M --bitrate 500k # # in another window on B, also as root... # ifconfig gr0 10.10.10.2 Now, on machine A you shold be able to ping machine B: $ ping 10.10.10.2 and you should see some output for each packet in the tunnel.py window if you used the -v option. Likewise, on machine B: $ ping 10.10.10.1 This now uses a carrier sense MAC, so you should be able to ssh between the machines, web browse, etc.