Defending telecom networks against the next Pokemon surprise

Most of the world has been surprised by the success of Pokémon Go, however, mobile network operators have been affected more than most companies. T-Mobile CEO John Legere tweeted that in four days the number of users on their network doubled while data usage quadrupled; recently released data shows that the average user spends more time playing Pokémon Go than using WhatsApp, Instagram or SnapChat.

Even though each of those users is generating under 10mb of data each hour, the cumulative effects are significant, especially as the game actively encourages users to move around thus moving the majority of this data away from wi-fi and onto mobile data provision. As specific locations are found to provide more opportunities, or feature rare Pokémon, it means that specific parts of the network will experience extra load, possibly even areas that previously saw low data use.

Add to this the thin margins in the mobile sector, the significant cost and regulations around upgrading infrastructure, and the fact that customer satisfaction is strongly linked with data capacity and the result is a situation where mobile operators could face serious threats to their business. On the other hand, providers looking to distinguish themselves have already spoken about offering packages that exclude Pokémon Go from data limits, prompting discussions about Net Neutrality rules. Finally, players will expect all this to ‘just work’ immediately, requiring responses outside the normal corporate planning timescale. Additionally the fact that there are some Pokémon unaccounted for suggests that they might be used for promotional events that could cause huge numbers of people to congregate and try and catch them, stressing the mobile networks and consequently damaging PR.

SpatialOS allows customers to simulate huge environments, including mobile networks on an international scale. ‘What If’ scenarios can be investigated demonstrating ‘emergent behaviour’ such as the recent Pokemon stampede, seen in the video below. 

With the spatial simulation capability of SpatialOS, mobile operators can experiment with potential impacts and derive plans of response, as well as understand what early warning signs they should look for, even before the game has been written. For example, when another Vaporeon appears in Central Park, a traditional model can show that people will congregate, however, the number, expected arrival rates and density are input values to the simulation, the modeller has to derive them from experiment, prior experience or guesswork. The truth or otherwise of these inputs is crucial to the validity of the model. SpatialOS can model all of the people in the city, and build behaviours based on what they perceive, this means that as they head towards the location and overload the local mobile capacity, they are also not going to receive tweets, and possibly move away once they realise what’s happening.

This video shows one of our simulations showing three independent technologies of cells serving the specific area.

This video shows one of our simulations showing three independent technologies of cells serving the specific area.

In this way SpatialOS can give insight into complex behaviours transcending the boundaries of individual layers in the simulation; in this case the interaction between network effect of people communicating via mobile phones, and the effect on the network of a large number of users in one place, something that would have to be specifically noticed and designed into traditional models. In SpatialOS simple behaviours, such as people reading social media about locations, heading towards them, and seeking out good reception, and the mobile infrastructure ‘breathing’ in response to load, can be combined to understand the richness of the interactions between the systems.

The simulation models the network and the subscribers, allowing for dynamic monitoring of cell load.

The simulation models the network and the subscribers, allowing for dynamic monitoring of cell load.

The responsiveness of data services are key to customer satisfaction, so it’s also vital to model the way data flows through the provider’s networks and out onto the internet as bottlenecks can exist anywhere in the infrastructure. The simulation can show where the network is at capacity, so potential solutions can be tested within the simulation resulting in either preemptive upgrades or actionable plans.

No one could have predicted the success of Pokemon Go, but it’s also unsurprising that occasionally a new app or game will take the world by storm. It’s important for any business to try and prepare for the future, and with SpatialOS’s large scale simulations it’s possible to explore situations that contain complex emergent behaviours, at a scale previously unimagined!