Bandwidth quantization for elastic optical networks with uniform traffic distribution

Most networks support continuous-rate channels, where the requests are being served by assigning them to the exact requested bandwidth, in other words, assigning them to the exact requested number of frequency slots (FSs) from the link bandwidth. Ultimately, this continuous supporting will lead to a loss of control in routing, billing, congestion, and admission. This problem may get exacerbated with the increasing complexity of the network design. Thus, an alternative network with small performance difference and less complexity design is needed. We propose to support all rates by implementing a number of equally spaced frequency levels with each level supporting a number of rates, and a way to solve this problem is to let the network implement certain rates. Then, all requests are forced to use the next higher bandwidth if the requested serving rate is not supported. We call this network a network with a quantized bandwidth. In this paper, we studied the effect of a quantized bandwidth on online routing and spectrum allocation (RSA) for elastic optical networks (EONs), in which the frequency slots must be assigned to each request contiguously and continuously. We studied the worst-case call distribution, which is the uniform distribution (maximum entropy probability distribution) because it has the highest randomness. Our results showed a small performance difference between EONs with quantized and unquantized bandwidths.