- This event has passed.
Special SDO Session – Low Latency Delivery of Metaverse Applications
The Metaverse Standards Forum organized Special SDO Session invites Forum members and Standards Development Organizations (SDOs) to present relevant ongoing or completed work to the MSF and the public.
Low Latency Delivery of Metaverse Applications
Low Latency, Low Loss, Scalable Throughput (L4S) and Low Latency DOCSIS (LLD) are two technologies aimed at dramatically reducing internet latency, especially in broadband networks. L4S introduces a new class of low-latency service using Explicit Congestion Notification (ECN) and scalable congestion control, enabling interactive applications—such as gaming, video conferencing, and VR—to perform with minimal delay and jitter even under network load. It allows co-existence with traditional traffic while maintaining high throughput. Complementing this, LLD is an enhancement to the DOCSIS (Data Over Cable Service Interface Specification) standard used in cable broadband. A similar roadmap has also been charted for L4S over Cellular networks. When used together, L4S and access network enablement will facilitate consistent, ultra-low latency performance across end-to-end internet paths.
The goal of this session is to provide Metaverse application developers an understanding of the steps network service providers are doing to enable low latency, and how developers can enable their applications to take advantage of these differentiated services.
Agenda
L4S Technology Brief and Implications for Metaverse Applications
The recently standardized L4S technology solves a fundamental problem in the Internet architecture: how can individual applications simultaneously achieve high throughput, low latency, and near-zero packet loss, without resorting to managed Quality of Service. This talk will give a brief overview of L4S technology and the current state of deployment. Further, it will discuss the applicability of L4S to metaverse applications and the available code components that can be leveraged by metaverse application developers.
Speaker: Greg White, Distinguished Technologist, CableLabs
Not Just Low Latency – How L4S Can Improve the Efficiency of Cellular Networks
End-to-end congestion control has proven to be problematic for volatile cellular networks, with wrong assumptions of packet loss, unnecessary retransmissions, and capacity-seeking flows counter-intuitively resulting in an inefficient use of available capacity. This talk explains the historic problem and how L4S can help solve it – whilst also supporting a choice for applications.
Speaker: Kevin Smith, Distinguished Engineer, Vodafone
L4S: The Next Steps Beyond TCP and AQM
While it brings a huge latency improvement to implement L4S by changing your TCP or QUIC based app to use the Prague CC and to add a DualPI2 AQM in your network bottleneck(s), you would still miss some further latency reduction opportunities. The latest Prague Congestion Controller doesn’t require that the network build a queue to control its rate, so why design networks and apps to still build queues? As networks will further evolve in avoiding queuing delays for L4S, this talk will introduce how to already fully optimize the latency of your L4S metaverse apps.
Speaker: Koen De Schepper, Research Principal, Nokia Bell Labs
Lessons From the First ISP to Deploy L4S at Scale
Comcast launched L4S in their network in a 2023 field trial and then full deployment in early 2025. With over 7M homes enabled with L4S and application partners like Apple, NVIDIA, Valve, and Meta, Comcast learned a great deal about the challenges in a large-scale deployment. They also have closely studied the performance benefits. This session will cover how their deployment works, what challenges they overcame, and what the end user and application benefits are.
Speaker: Jason Livingood, Vice President, Comcast
