The Internet is changing as a result of extensive integration of wireless networks and mobile computers into already-existing infrastructures. This transformation, however, brings forth inherent challenges: bandwidth constraints, latency spikes, error-prone connections, and frequent disconnections, all exacerbating network losses during handoffs. Compounding these issues, TCP performance degrades in wireless settings. Our paper tackles these complexities by presenting a novel approach to improve end-to-end reliable transport in mobile wireless environments. Based on a novel methodology based on the normalized delay gradient (NDG) loss predictor function, we enable mobile hosts and base stations to dynamically modify sender window sizes in response to different types of losses, such as congestion or handoff-induced losses. This adaptive strategy preserves TCP’s integrity end-to-end while optimizing network-layer operations. Using stochastic modeling, we thoroughly examine crucial parameters, including handoff call rates, blocking probabilities, loss attribution (congestion versus handoff), sender window dynamics, router queue lengths, throughput metrics, and impacts of congestion and handoff-induced losses. Our findings provide avenues to mitigate handoff-related call drops and provide accurate, forward-looking solutions for mobile wireless systems.