Designing Stadium Wi-Fi for successful offload

amex-stadiumPeople often tell me that you couldn't use cellular small cells to cover a sports stadium – other more expensive cellular systems are required. At the same time, we've seen Wi-Fi successfully deployed in these same stadium environments. ATT carried 1.5TBytes of data over Wi-FI at the SuperBowl, three times that of their cellular traffic. Sami Susiaho explained at BBWF how The Cloud (a UK Wi-Fi network) designed and deployed a Wi-Fi solution at Brighton and Hove Albion stadium during summer 2014.

The starting point is to decide on dimensioning

Industry views differ widely about how many users would concurrently access Wi-Fi in a stadium. Anything from 20% to 200% is mentioned. Usage patterns will vary in different areas, with the press zone being much more demanding that the average stadium seat. This means that different design rules are required for different areas.

For the main stadium bowl, The Cloud designed for 50% concurrent access for the maximum 30,000 crowd, connecting 80 Wi-Fi access points using 1km of fibre and 9km of CAT6 ethernet cable.

Each access point can handle up to 250 concurrent users. Tightly focussed beams were used to segregate seating blocks, splitting these into distinct coverage sectors.

Keep it simple

To simplify the design, the older 802.11b standard wasn't used/supported, VoIP was blocked and a maximum of 3 SSIDs assigned. Unlike a cellular system, there's no handoff as you move around the stadium – you'd need to reconnect and create another session. During peak usage, almost everyone is sitting down rather than moving about (if you ignore those jumping up and down on their seats).

Both Wi-Fi spectrum bands at 2.4GHz and 5GHz were used, with devices capable of the higher frequency prioritised to use it. 56% of clients used the 5GHz band, which has much more spectrum and many more channels available. The different propagation characteristics mean there are different coverage footprints, so planners are actually designing two networks rather than one.

The side lobes on the 5GHz coverage footprint were massive, limiting the number of Wi-Fi access points that could be deployed.

Accurate predictions from the radio planning tool

Measurements taken inside the stadium during live operation aligned closely with predicted performance. The physical presence of large numbers of people make a huge difference to how the network performs – bodies absorb some of the low level signals, especially at 2.4GHz more than 5GHz – giving relatively few short windows of real-time operation when active measurements can be made.

It is during these times that different parameters and tweaks are made to improve throughput and performance. This limits the number of measure/analyse/review cycles for parameters changes and performance tweaks. The stadium is only used for a few hours at a time, with some 150 hours in total throughout the year.

The coverage pattern below shows that the centre of the pitch has (as expected) no signal

Stadium-Wi-Fi

Average 17Mbps potential downlink throughput for 90% of users

The way performance was quoted didn't state actual throughput, but instead the potential downlink available to users at any time. When the system was first installed, average potential performance was in the low Mbps. With expertise and effort, this has been raised to 17Mbps for the 90th percentile (ie in 90% of the time, a user can expect to request and receive 17Mbps).

Statistics from a recent event:

Total attendees 24,569

...Of which 36% had unique Wi-Fi devices "seen" by the Access Points (8,844)

...Of which 35% actively authenticated on their network (3,095)

Total 320,187 minutes of use by 4,793 sessions.

Over 50% of devices were Apple IOS, less than 20% Android.

Blueprint for success

  • Take control of which devices associate to which Wi-Fi Access Point
  • Design 2.4GHz and 5GHz separately
  • Carefully design coverage areas to match the use case: moving/stationary, roaming
  • Simulate the effect of people and how bodies attenuate the low power RF
  • Expect only a few hours of usage per review cycle and 150 hours of use per year
  • Design, Deploy, Measure, Evolve

Treat these projects as OPEX not CAPEX. They are an ongoing service and design project rather than a one-off installation.

It's not about the equipment, it's much more about the design.

 I suspect many of these points would relate equally to Small Cell and DAS deployments in high profile locations.

TheCloud is UK based Wi-Fi network operator and aggregator owned by BSkyB, serving over 10 million users through 22,000 hotspots.

Hits : 3448
  • 4

    more

    Residential

    Residential

    A significant number of users continue to report poor mobile coverage in their homes. There will always be areas which are uneconomic for mobile operator to reach. They range from rural areas

    ...
  • 4

    more

    Enterprise

    Enterprise

    The term Enterprise addresses any non-residential in-building including hotels, convention centres, transport hubs, offices, hospitals and retail outlets. It's not just intended for businesses to

    ...
  • 4

    more

    Urban

    Urban

    Urban small cells (sometimes also named metrocells) are compact and discrete mobile phone basestations, unobstrusively located in urban areas. They can be mounted on lampposts, positioned on the

    ...
  • 4

    more

    Rural

    Rural

    A rural small cell is a low power mobile phone base station designed to bring mobile phone service to small pockets of population in remote rural areas. These could be hamlets, small villages or

    ...
Categories
Backhaul Timing and Sync Chipsets Wi-Fi LTE TDD Regional

Popular Categories

Follow us on...

footer-logo

Search