I have heard that non overlapping channels (1,6,11) in a G radio do overlap at power level set to the maximum.
Is this true...? Pls neglect if this is not the case.
This is true.
Please read the following whitepaper. You have the details on ajacent interference on Page 4 with a diagram on Page 5.
While study for the CWNA I found several references to ERP-OFDM being more robust them HR-DSSS and there by propagating further (at the same Tx Freq. and same data rate) and providing a stronger RSSI to the receiver (if both transmissions are the same distance) while having the same EIRP at the Tx Ant (a slight decrease in free space path loss). It may not be much greater but it is still there. So, with a more robust signal the sidebands will experience less attenuation and possible greater bleed-over into other channels.
[i]At 11 MHz from the center, the transmitter energy level is only 20 dB below the maximum (as opposed to 35 dB for 802.11b), and at 22 MHz away, the energy is only about 30 dB below (as opposed to 50 db for 802.11b). Even as far out as 40 MHz, the energy is still only 40 dB below the maximum.
The spectral efficiency (that is, the way in which the frequencies surrounding the center of the channel are used) for the orthogonal frequency division multiplexing (OFDM) modulation used in 802.11g devices is much worse than that for the complementary code keying (CCK) modulation used in 802.11b devices. Figure 7 shows the transmitter specification for 802.11g. [/i]
My understanding is that it is not just power but distance that needs to be considered when looking at interference between non-overlapping channels (co-location/proximity).
[i]Although one would not expect any interference between non-overlapping channels, experiments showed that there is interference when the transmitters or receivers are in close proximity. The non-overlapping nature of the channels is true when the antennas are beyond a certain distance from each other.[/i]
The 802.11a clause 17 OFDM sidebands do overlap but are attenuated enough that they do not breach the threshold and are not of concern.
You are correct in the sense interference would decrease on increasing the distance between the APs on non overlapping channels.
O/P power is directly proportional to distance. More the power more the distance the coverage would be. Hence the theory.