6.3 Simulation Results and Evaluation
In this section, we evaluate and compare the performance of FSA with SA and CSA in GloMoSim (Zeng et al. 1998). We also introduce a perfect detection-based scheme, IDEAL, as the baseline for comparison. The IDEAL scheme is the same as FSA, except that all the ACKs in IDEAL are 100% reliable and the detection judgement is 100% precise. As our focus is on the efficiency of different coordination schemes given the same candidate set and the corresponding forwarding priorities, we use an existing candidate selection algorithm based on a node's geographic locations and adopt the local metric expected one-hop throughput (EOT) (Zeng et al. 2007) to select candidates. We will elaborate this local metric in Chapter 8.
Because the existing popular network simulators, such as NS-2 (McCanne and Floyd 1997), OPNET (www.opnet.com), GloMoSim (Zeng et al. 1998), have not implemented the PHY layer's function like energy integration or matched filter module currently, we have done some modification to the PHY layer in GloMoSim and make the detection judgment based on the following probability model. We define the CCA error floor (Zhen et al. 2008) at the optimal threshold, which can be achieved by equating 1 − Pd and Pfa where Pd is detection possibility and Pfa is the false alarm possibility. Then the CCA error floor for ED-based CCA and PD-based CCA can be expressed in terms of the Q function (Kay 1998):
We define the following performance metrics.