The main idea behind “diversity” is to provide different replicas of the transmitted signal to the receiver. If these different replicas fade independently, it is less probable to have all copies of the transmitted signal in deep fade simultaneously. Therefore, the receiver can reliably decode the transmitted signal using these received signals. This can be done, for example, by picking the signal with the highest SNR or by combining the multiple received signals. As a result, the probability of outage will be lower in the case that we receive multiple replicas of the signal using diversity. To define diversity quantitatively, we use the relationship between the received SNR, denoted by γ , and the probability of error, denoted by Pe. A tractable definition of the diversity, or diversity gain, is

 Gd = − limγ→∞log(Pe)/log(γ )

 where Pe is the error probability at an SNR equal to γ . In other words, diversity is the slope of the error probability curve in terms of the received SNR in a log-log scale. There are two important issues related to the concept of diversity. One is how to provide the replicas of the transmitted signal at the receiver with the lowest possible consumption of the power, bandwidth, decoding complexity and other resources. The second issue is how to use these replicas of the transmitted signal at the receiver in order to have the highest reduction in the probability of error.