Hopefully this isn't too far off topic. I've just completed testing of
an OFDM modem using IP over DVB-T2. It uses an SDR transmitter and a
commercial DVB-T2 receiver to implement the RFC 4326 Unidirectional
Lightweight Encapsulation (ULE) for Transmission of IP Datagrams over an
MPEG-2 Transport Stream (TS). ULE is supported in the Linux kernel for
DVB receivers.
https://tools.ietf.org/html/rfc4326
It's a full-duplex modem capable of up to 50 Mbps (in both directions)
in an 8 MHz bandwidth. The current test bed consists of an Ettus B200
SDR transmitter, PCTV 292e DVB-T2 USB receiver, Kuhne down converters
for 13cm and 9cm, Microlab BK-26N diplexer and RFSpace TSA600 Vivaldi
antenna.
http://www.w6rz.net/IMG_0119.jpg
http://www.w6rz.net/traceroute.png
The transmitter is based on the DVB-T2 transmitter in GNU Radio and uses
this OOT module for the ULE protocol.
https://github.com/drmpeg/gr-ule
To reduce the latency, I've merged the DVB-T2 blocks to avoid having so
many buffers between blocks.
https://github.com/drmpeg/gr-dvbt2ll
Current test frequencies are 2305 and 3429 MHz with an 8 MHz bandwidth.
The bit-rate is 28.6 Mbps (symmetrical).
It's intended to replace commercial WiFi equipment for amateur WAN
interlinks.
Advantages:
1) Full-duplex. Adding power amplifiers, preamps, diplexers/duplexers is
easy.
2) Frequency agile. Can work on any band above 420 MHz. 70cm through 5cm
direct TX from the SDR and millimeter frequencies with an up-converter.
3) Bandwidth agile. 5, 6 ,7 and 8 MHz bandwidths.
4) May be legal on 70cm. Although I haven't implemented it yet, a small
portion of the bandwidth can be used to send a low-rate video stream
(for example, a still picture of your call sign for ID). This would
classify the emission as digital ATV, not data.
Disadvantages:
1) Latency is a bit high. It's currently 100 ms (200 ms round trip).
This is a function of buffering in GNU Radio and the USB 3.0 connection
to the SDR. An FPGA implementation of DVB-T2 and a different SDR
architecture could solve this.
2) Cost. It's difficult to compete with commercial WiFi equipment.
However, lower cost components can be used instead of the "Cadillac"
test bed I constructed. For example, a ADALM-PLUTO at $99 could be used
instead of the Ettus B200 for transmit. Two antennas instead of a
diplexer and lower cost down-converters than the Kuhne units.
3) Requires a Linux computer to run it. An Odroid XU4 may be adequate,
but I haven't tested it.
73,
Ron W6RZ