The Pokémon Go effect on the network

Imagine sitting courtside at game seven of the NBA finals without having to pay the reported $99,000 someone spent on two tickets last month. Or imagine watching that same game in the stadium, and live statistics pop up on your connected glasses that show a certain player is approaching a triple-double—without having to take your eyes off the action.

If you think either seem unrealistic, you probably didn’t think you’d be bumping into people chasing virtual Pokémon around the streets, did you?

+ Also on Network World: Amazon CTO says cloud can help crashing Pokemon Go +

For many in the virtual reality (VR) and augmented reality (AR) spaces, sport is the end game: streaming live with a VR device can take the fan to places—in the stadium or on the field—they never thought possible, while including the freedom to look where they want, when they want. And AR would augment being at an event with up-to-the-minute information without having to peek down at their phone; they’d be able to keep their eyes up on the play.

A number of specialist VR companies have already effectively trailed and streamed at least a version of live-sports VR. In fact, the technology is set to trial during the upcoming Rio Olympics, with the Olympic Broadcasting Service promising to broadcast high-definition footage of the opening and closing ceremony in virtual reality in addition to one event per day—though full details of how this will appear have yet to be fully revealed.

But are we putting the cart before the horse here in a way? Do we have the underlying network infrastructure in place to make the VR and AR experience as seamless as possible for mass audiences? That is the critical piece to ensure success from the get-go, and ultimately that experience comes down to bandwidth and latency. And with VR and AR adding to an already-jammed over-the-top crowd of applications that need seamless experiences to survive, we’ll hit the theoretical limit of Shannon’s Limit sooner than we think.

Can the network handle the demand?

Let’s take AR and the Pokémon Go effect on the network. It’s clear that the technology is viable given the explosion in popularity of the game. We all know things can “go viral.” But can today’s network really support the needed scale as the demand per user and pre application also “goes viral?” Critical to this game, VR and AR in general is constant high-quality connectivity. Lose that, and you get a bunch of angry Pokémon chasers. And this is just one game. Imagine we constrain the bandwidth further with new AR games and applications on top of the demands we’re already placing on our 4G networks.

It’s clear that 5G is key to ramping this up. Bringing the bandwidth to the user from a wireless perspective is important in relation to the off-loading of compute-intensive tasks from the device to the cloud. This will benefit the device by reducing power consumption, and it will benefit the user by taking advantage of the compute power of cloud.

We tend to forget that each device that enables fully-immersive VR and AR experiences—sensors, cameras, microphones, smartphones, glasses and others that have yet to be dreamed up—are all connected. They’re part of this Internet of Things (IoT) era that we’re firmly in. And on the scale of bandwidth rapaciousness, VR and AR are on the hungrier end of the IoT scale.

With VR, some camera rigs could require 100 Gbps of contribution bandwidth. Most networks are only now transitioning to be 1 Gbps capable at the edge with 100 Gbps at the core of the network. Should VR hit critical mass quickly, that bandwidth could be soaked up by sports fans and gamers, leaving no capacity for other services.

Latency is also key. That’s because even small delays in routing the round-trip signal can have a major adverse effect on the VR experience, potentially causing motion sickness for the viewer; it’s disorienting for the viewer to turn his head at a certain speed and not have the landscape move at the same pace. While that’s not as much of an issue with VR content that is downloaded locally, it can be a major problem if the content is being streamed and the latency and bandwidth capabilities aren’t at optimal levels.

What the network must become for this to work

The limitations for VR and AR right now are due to the underlying ecosystem, and we must get compute, store and connect all working seamlessly and closer to the user. No single company can solve this. It involves leveraging software-defined networking (SDN) to evolve the network into an on-demand service with the capability to ramp resources up or down as required.

The network must become smarter and programmable. All of the resources in the infrastructure need to be orchestrated together to provide the desired end-user experience. The network needs to become responsive to applications. It needs to know the times traffic spikes are likely to occur—for example, during the opening ceremony of the Olympics for VR, the lunchtime and weekend for Pokémon Go, the local event that everyone wants to stream—and be able to react and respond in real time by providing the necessary connectivity and capacity.

The answer to the problem of increased demand on the network is to flip that phrase around and evolve to what can be called network on demand. Network topology, connectivity, service class and quality of service all need to be on-demand services that can be customized to suit the needs of the end user.

When that underlying network is ready, only then can we begin to cater for increased demand. And when that happens, the future of VR and AR looks bright.