KM & Phil tested point to point WiFi with parabolic antennas and 500mW amps at each end. Range was at least 8.5km! Amazingly one end connected without its amp - and the long white omni antenna connected as well! The amps use a 3.45mm connector, maplin order code L45AY. Test in Iceland covered Holmur to cafe on mountain easily (16km).
Less directional antenna
Experiment on 11/6/11 by KM using:
small usb-wifi adaptor -> sma-N adaptor -> WiFi-link Yagi WLY-2450 -> onmiWLO-2450-08 -> N-sma-> EnGenius ECB1220R.
Signal was -68dB at 1.5km ! ie useable. angle of Yagi not too sensitive.
high power 868MHz units  are best yet - tested to 500m range in forest. Also over 50m when buried in the forest (at 200mW). Out model is discovered by X-CTU as XBP08-DP Xbee-pro 868 single channel tunable from 1-300mW
868MHz 12dBm worked across our building - with only pcb antenna! this gives us confidence in this type of setup for many deployments.
[] seems to operate at only 2.4kbit/s at around 868.14MHz.
Tests OK with one node 40cm under soil... - but needs a better antenna and wet soil test.
We have now frozen work on these particular devices.
xbee pro 868: It turns out these units, although powerful, can only run at 10% of their bandwidth due to legal air-time limits.
ETSI docs say transmitter needs to stop for 100ms between transmissions (which can not be longer than 4s).
Does that mean hopping network - where transmitting node will wait a while (for ack/tcp etc) is within specs?
Short Burst Data
This satellite-messaging system (5m per 50 char message) looks ideal for low data links and we have investiogated it as a backup channel for remote systems. RockBlock hardware from rock7.
cables and connectors
remember RF cable such as RG58 or RG174 looses about 1 or 2 dB/m respectively (at 2.4GHz) this could be important for long range coms and GPS cables.