SpectrumTalk

The independent blog on spectrum policy issues
that welcomes your input on the key policy issues of the day.

Our focus is the relationship between spectrum policy
and technical innnovation.

Can Cognitive Radio Technology Help Solve Some Difficult Spectrum Management Issues by Creating "Virtual Guardbands"?

wcm-logo
In the April 2011 issue of IEEE Wireless Communications magazine is an article by your blogger with the above title. It explores an idea that first came up during the ill fated AWS-3 deliberations: are there special cases where the “hidden node problem” can be solved?

The hidden node problem refers to the problem of protecting receivers from interference in a cognitive radio system when you can’t explicitly detect the receiver’s presence. This is the crux of the TV whitespace problem since TV receivers can not announce their presence. (In that case the use of cognitive radio detectors 20+ dB more sensitive than TV receivers can be used to solve the problem although FCC chose the more conservative approach of geolocation for initial implementation.)

The instant article considers the case of a full duplex mobile transmitter with known frequency offset, as in the case of the ubiquitous cellphone. It points out that in this case the hidden node is no longer “hidden” since it can be readily detected at short distances by the uplink signal and that the receiver frequency can also be calculated by adding the offset to the detected uplink frequency.

This is a special case of Preston Marshall’s general observation that while cognitive radios are best known for finding empty spectrum that can be used without cochannel interference, if you know the location and characteristics of nearby receivers you can also select frequencies that will not result in interference from adjacent channel or intermodulation mechanisms in receivers with limited rejection capability. Pres points out that receiver ejection capability is generally expensive because it requires on precise analog components, whereas cognitive radio technology is mostly digital with little marginal cost in the long run.

In the AWS-3 proceeding, T-Mobile, the adjacent channel licensee in most places, claimed that adjacent channel interference to their FDD downlink was inevitable if the adjacent AWS-3 were used for either uplinks or TDD. Since we now know that T-Mobile was “on the auction block” during this deliberation, they had every incentive to maximize their sale value by using a massive legal effort for both trying to stop a new entrant and removing any risk of interference from concerns of potential purchasers. One of the parties in the AWS-3 proceeding mentioned a variant of virtual guardbands in comments, but this was promptly dismissed by the cellular mainstream as speculative.

But while there might be some slight question of how well TV whitespace devices can detect a TV signal from 50 miles away that provides marginal reception to TV sets using antennas near the devices, detecting a cellphone uplink signal nearby - the only place where adjacent channel or intermodulation interference is a threat - is trivial by comparison.

Industry loves to demand guardbands: it is a proven technology to prevent interference and also limits competition. But guardbands also have a real cost in our spectrum dependent economy that is always searching for more spectrum. Virtual guardbands can be a realistic way to use today’s technology to avoid interference while allowing more intense spectrum use at the boundaries of FDD bands.

FCC has never resolved the AWS-3 issue, they just told M2Z Networks that no answer was coming soon. AWS-3 continues to lie fallow as we beat the bushes to find the 500 MHz of new spectrum demanded by the National Broadband Plan. - there are no easy answers. Maybe with technologies such as virtual guardbands we can put AWS-3 to use serving society and the economy. With AT&T and T-Mobile now before the Commission begging for permission to merge, maybe they can more more flexible on the AWS-3 issue as a merger condition.
Comments