There is a lot of literature on this subject. These should get you started:
http://ecewp.ece.wpi.edu/wordpress/wireless/files/2011/11/ece4305_lab4.pdf
http://airccse.org/journal/ijngn/papers/3211ijngn03.pdf
The second article gives a nice graphic of the various techniques, and then describes pros and cons of each. You might get research ideas from that. There are plenty of aspects still to be optimized.
Well, there are various techniques used in CR for spectrum sensing but all depends on the amount of information available for SU about PU signal, the more information the better (probably more complex) detector you can construct.
For instance, the simplest form is the energy detector which does not need any a prior information about the PU signal, it simply samples the environment and decide based on a threshold if the signal exists or not. This detector can fail in low SNR PU signals.
The cyclostationary feature detector can resolve this problem in Low SNR but it's more complex and it requires a cyclic signal from PU with a known period.
The matched filter detector and entropy based detector are other examples.
those are single detectors. but if you are dealing with a network and not a single detector it's highly recommended to cooperate between SUs in order to improve decision accuracy because the cooperation can increase the diversity of the decisions and then can overcome problems at some nodes like shadowing, deep fading and hidden terminal. to make a final deicison from the measurements you have to employ a fusion technique.
Thanks.
- Badges
- Science topic
- Similar topics
- Computer Science
- Research Topics