POSTER: DGSP — Demand-Grouping and Spectrum-Partitioning-Based Spectrum Allocation in Elastic Optical Networks

In elastic optical networks (EONs), spectrum fragmentation occurs when available spectrum resources become split into small, non-contiguous blocks, resulting in inefficient spectrum usage. Traditional methods, which allocate separate, dedicated partitions for each bandwidth demand type, often lead to an underutilized spectrum. Dedicated partitioning limits available slots per demand type, significantly increasing the likelihood of blocking requests, particularly those with higher bandwidth demands—even if sufficient spectrum slots are free in other partitions. To address this challenge, this paper proposes DGSP, a bandwidth demand grouping and spectrum partitioning scheme designed to reduce connection blocking. DGSP utilizes an integer linear programming (ILP) model to optimally group demands based on their slot requirements and determine equitable partition sizes. Within each partition, spectrum allocation is done using the first-last fit (FLF) policy. Simulation studies conducted on two networks confirm that DGSP effectively reduces the blocking probability, achieving improvements of up to 97.51% compared to conventional dedicated partitioning approaches.