An adaptive congestion control algorithm for improving Transmission Control Protocol performance over cellular-to-cloud networks

In recent years, significant advancements have been made in enhancing congestion control algorithms (CCAs) within the Transmission Control Protocol (TCP) for end-to-end communication in millimeter-wave (mmWave) cellular networks. However, TCP often struggles to effectively utilize available bandwidth in mmWave-based cellular-to-cloud networks due to round-trip time (RTT) constraints, resulting in suboptimal throughput and latency that negatively impact cellular-to-cloud applications. To address this limitation, we propose an adaptive CCA, MRVHS-based CCA, which integrates maximum segment size (MSS) and RTT variations into a comprehensive CCA framework. MRVHS is implemented and evaluated using the ns-3 network simulator, with its performance compared against well-established TCP variants, including NewReno, HighSpeed, CUBIC, bottleneck bandwidth and RTT (BBR), and fuzzy logic-based TCP (FB-TCP) in cellular-to-cloud networks. Simulation results indicate that MRVHS consistently achieves higher average throughput while maintaining low latency across varying packet error rate (PER) levels. Notably, MRVHS significantly outperforms HighSpeed and CUBIC at the highest PER, achieving a 4.75% improvement over BBR in high-PER scenarios. Moreover, MRVHS demonstrates substantial throughput gains in medium- and low-PER conditions, consistently surpassing the benchmark TCP protocols. Furthermore, MRVHS demonstrates substantial throughput gains under medium and low PER conditions, surpassing the benchmark TCP protocols.