A novel fragmentation metric and fragmentation-aware adaptive routing and spectrum allocation algorithm in elastic optical network

Next-generation elastic optical networks are being developed to accommodate the increasing demands of heterogeneous traffic for 5G/6G communication. The arrival and departure of this real-time traffic are inherently dynamic. Such dynamic traffic requests, combined with varying bandwidth requirements, lead to spectral fragmentation within the links. Link fragmentation refers to the presence of non-contiguous available spectrum slices within these links. An increase in the number of link fragments diminishes the width of the available frequency slots, which may not meet spectrum constraints, thereby resulting in higher blocking rates within the network. Consequently, it is essential to assess link spectral fragmentation to implement an effective spectrum management strategy. Various metrics for measuring fragmentation in links have been proposed in the literature, including external fragmentation metric (EFM), entropy-based fragmentation metric (EBFM), root mean square fragmentation metric (RMSFM), and golden fragmentation metric (GFM). This paper highlights the limitations of these existing metrics and introduces a novel spectrum slice-based fragmentation metric (SSFM). To evaluate the effectiveness of SSFM, in comparison to EFM, EBFM, RMSFM and GFM, a fragmentation status aware adaptive routing and spectrum allocation (FSA-RSA) algorithm has been developed. Simulations are conducted on German, USNET and Telecom Italia (TI) network topologies. To evaluate the effectiveness of SSFM in relation to other fragmentation metrics, we have employed the following measures: Execution Time (ET), Computational Overhead Ratio (COR), Memory Overhead (MO) and we have evaluated the following performance metrics as a quantitative tool for measuring the effectiveness of the proposed FSA-RSA (SSFM) algorithm against benchmark algorithms: throughput, Request Blocking Probability (RBP), Fractional Spectrum Utilization (FSU), and Link Fragmentation Metric (LFM). Among these, throughput and execution time are regarded as the primary performance metrics. The findings suggest that the FSA-RSA (SSFM) algorithm enhances network performance, yielding an average minimum throughput increase of 3.5%, 0.12% and 108% when compared to the FSA-RSA (EBFM), FSA-RSA (EFM) and FSA-RSA (GFM) algorithms, respectively. However, it exhibits a throughput that is 0.0049% lower than that of the FSA-RSA (RMSFM) algorithm, a shortfall that can be offset by the reduced execution time and memory overhead associated with SSFM in comparison to RMSFM.