Although a cognitive radio solution looks very promissing, especially for non mission and life-critical M2M communications, e.g. metering and monitoring data, applications that involve tightly coupled monitoring and control loops may require very high reliability and very low latency (e.g. ~milisecond for safety-of-life ITS applications, or some applications in smart grids, such as phasor monitoring) which initially seem unachievable using opportunistically acquired spectrum.
Two issues arise
-1. M2M communication using cognitive radio is interuptable (this can happen when a primary system shows up in the band)
2. Even if this problem coul be fixed, e.g. because primary sytems spectral occupancy is predictable in space and time (e.g. using a geolocation database approach), then theres is the issue of interference due to other cognitive radios/secondary users operating in the band.
There is, in my view, no straighforward solution here. One option could be that a particular bands is exclusively used/dedicated for secondary use of mission/life/safety critical M2M use, similiar to the recent approach that is being adopted for the MBABN (Medical Body Area Networks) in the United States, and is then managed using a database approach. Another, more amitious approach woul be to rely on the ability of cognitive radios to switch channel/spectrum quickly and intelligently in a way that QoS is maintined despite intermitency in spectrum availabilit, previous work on channel switching in IEEE 802.22 networks (mainly simulations though) suggest this is possible, at least to satisfy QoS requirements of human-driven traffic. A final option is that M2M sytems would rely on a mix of licensed, license-exempt and opportunitically acquired spectrum.
See also my forthcoming paper in the IEEE Network
- Cognitive machine-to-machine communications: Visions and potentials for the smart grid, with Yan Zhang, Maziar Nekovee et al.
