With spectrum sensing you don’t pick a clear channel and stick with it indefinitely, you keep scanning every few seconds and hop to a new channel as needed. This allows you to not only avoid interfering with other devices but also always be on the channel with the best RF characteristics. The overall mechanism would be very similar to frequency hopping in GSM. When the user device is powered on it scans through all channels and associates with the base-station that it can receive the best. If it doesn’t find any base stations, it will continue to periodically rescan until it does. Channel hopping will be controlled by the base station. When it decides there is a need to change channel it will first broadcast a message to all user devices informing them of the new channel. Even if this message is missed by a particular device, after a period of no control messages being received it will re-enter scanning mode and eventually rediscover the base-station (by eventually, I mean some fraction of a second). A primary factor in the success of this strategy is that user devices will have to be limited to no more power than devices like wireless mikes, so that you will never have a situation where the base-station cannot “hear” such devices in the proximity of the White Spaces user device. In spectrum starved markets, dynamic power control can also be employed to reshape the service area on the fly as wireless mikes and whatnot are detected — sucks for users near those wireless mikes, but at least the rest of the network is unaffected. These techniques have all been used to various degrees in digital radios for a very long time, so we are not talking about anything hard or costly to implement in prospective White Spaces devices (although standardizing things for guaranteed interoperability between vendors could take years).