sasirekha_spectrum_sensing.ppt

Editor's Notes

• #9: Cognitive Radio Ad Hoc Networks comprise of Cognitive Radios (CRs) also referred to as Secondary Users (SUs) connected in an adhoc manner. They could be of centralized or de-centralized architectures. The primary feature of a CR is Dynamic Spectrum Access (DSA). The major components associated with DSA are 1. Spectrum Sensing 2. Spectrum Allocation 3. Spectrum Mobility. Spectrum sensing is thus an important functionality of DSA. Spectrum sensing, i.e. identification of the spectrum occupancy by the Primary User (PU) can be done either in standalone or collaborative manner. Collaborative spectrum sensing combats hidden node and fading conditions of the CR.This collaborative spectrum sensing can be applied to Cognitive Radio Ad Hoc Networks (CRAHNs).
• #10: The CRAHNs when applied to emergency situations like military and public safet (Disaster management) networks, they can be called emergency CRAHNs. Their system model could be of nomadic mobility, where each group moves from one place to another for their operation. The internode movement can be random. The group can be associated with an average Group SNR. This group senses the presence or absence of PUs collaboratively.
• #11: This literature survey helped us to formulate the requirements of Spectrum Sensing in Emergency CRAHNs, The spectrum sensing methodology needs to be accurate,Resource efficient,Low latency in the delivery of packets,Adaptive to varying number of SUs Adaptive to varying SNR conditions, Fairness in energy consumption because of the battery operation of the devices.The method should also show. Resilience to data falsification attacks (Byzantine attacks). Here the attacker sends false information on the spectral occupancy. Spectrum sensing being a complex issue with several dependant parametrs across the various layers of the protocol stack, poses a challenge to fulfill all the above requiremnts. The spectrum sensing problem is a complex one depending on parameters of physical and link layers. The optimization problem can be looked upon in 2 levels. Fisrt one is the local optimization where the SNR, Threshold, channel type and the sensing mechanism play a vital role in determining of local sensing performance. The performance is measured in terms of probability of detect and prob. of false alarm. The time scheduled for sensing And the frequency of sensing also define the local performance. The data from different SUs is combined to form the reliable decisions. The fused data accuracy is mesured in terms of Risk which is nothing but the weighted sum of Prob. Of detection and prob. Of false alarm of the fused data. The risk depends on the fusion rule applied, the PU usage pattern, number of SUS used for fusion. These parametrs can be mapped to the following framework.
• #12: 11
• #14: In game theory there are actions and their payoff. One shot game is played only once, it is like an interaction in train we meet the person only once. The iterative game is played several times, utility is accumulated. Like an interaction with neighbours, can’t make hasty decisions, we need to to have a long term relationship. So actions will be different. Finally we have the evolutionary games where you change the strategy according to your belief, and external factors. This belief is built over time. We had tried out two approaches, iterative and evolutionary. One more point to be noted is that in any public good game say road building, we need a reward or penalty to make people contribute. Else it will fail. In our approaches we tried both reward and penalty.