Vector quantization based QoS evaluation in cognitive radio networks

In this paper, we attempt to characterize the QoS that secondary users can expect in a cognitive radio network. Using power control as a black-box, we propose a method that can help us evaluate the QoS for any given power vector based on past observations. To that end, we first define a k-dimensional QoS space where each point in that space characterizes the expected QoS. We show how the operating condition of the system maps to a point in the QoS space, the quality of which is given by the corresponding QoS index. To deal with the real-valued QoS space, we use vector quantization to partition the space into finite number of regions each of which is represented by one QoS index. We argue that any operating condition of the system can be mapped to one of the pre-computed QoS indices using a simple look-up in O(log n) time- thus avoiding any cumbersome computation for QoS evaluation. Using simulations, we illustrate how a 2-dimensional QoS space can be constructed. We choose capacity as the QoS metric and show what the expected capacity would be for a given power vector.