Waseda University AI model achieves 87 percent error-free 5G multicast broadcasting

This development addresses the critical lack of retransmission capabilities in one-way wireless delivery, which has historically caused video stuttering in multicast environments. By shifting from reactive retransmission to predictive adjustment, 5G multicast broadcasting AI allows mobile networks to function as viable alternatives to costly fiber-optic retrofitting in older infrastructure. For the broader streaming ecosystem, this provides a spectrally efficient path to deliver linear television and emergency alerts to mass audiences without congesting unicast bandwidth. Watch for whether this lightweight architecture is integrated into upcoming 3GPP standards for next-generation broadcast-broadband convergence.

The 3GPP Release 17 specification formally introduced Multicast/Broadcast Services (MBS) as a standardized feature for 5G networks, establishing the technical foundation on which Waseda University's predictive model operates. In March 2025, the 3GPP confirmed that Release 18 work items include enhancements to MBS for improved spectral efficiency and expanded service scenarios, building on the baseline multicast framework that Release 17 defined. South Korea's KT Corporation conducted one of the earliest commercial 5G broadcast trials, demonstrating multicast delivery of live sports content to multiple simultaneous users on a single cell in September 2024, validating the network-side infrastructure that lightweight AI models like Waseda's are designed to complement. On the regulatory and business side, the European Broadcasting Union has been actively promoting 5G broadcast as a complement to traditional DVB-T2 terrestrial delivery. In November 2024, the EBU published a position paper urging EU member states to reserve spectrum in the 470-694 MHz band for hybrid broadcast-broadband services, arguing that multicast delivery over 5G could reduce the carbon footprint of linear TV distribution by up to 40% compared to unicast streaming at scale. Meanwhile, Qualcomm has positioned its Snapdragon X75 modem as the first chipset to natively support 5G NR MBS reception, announcing in February 2025 that the platform enables simultaneous multicast and unicast operation without additional hardware, which aligns with Waseda's emphasis on running their model on consumer-grade smartphone chipsets rather than specialized broadcast receivers. From a technical benchmarking perspective, the challenge Waseda's team addresses is well documented in independent research. A 2024 study published by the IEEE Communications Society found that conventional 5G multicast implementations using RLC acknowledged mode experienced packet loss rates exceeding 60% in high-mobility scenarios, confirming the severe degradation that occurs when retransmission is unavailable. The Waseda model's 87% error-free rate represents a meaningful improvement, though it remains below the 99.9% reliability threshold that 3GPP specifies for ultra-reliable low-latency communications. Competing approaches from Nokia and Huawei have focused on network-side AI for multicast scheduling optimization rather than device-side prediction, with Nokia presenting a reinforcement-learning-based MBS scheduler at MWC 2025 that reduced rebuffering events by 45% in simulated urban environments, suggesting a complementary rather than directly competing research direction.