Dubai Airport is looking to leverage the huge volumes of operational data it has to start driving better customer service across the airport, from the security queue times all the way to the toilets.
Michael Ibbitson is the former CIO at Gatwick Airport, which has been using big data analytics tools from Splunk in a bid to reach zero-queuing times through security. Now in his new role as EVP for technology and infrastructure at Dubai Airports, the executive is trying to take those practices and scale them up for the larger airport.
Dubai airport processes 90 million passengers a year, making it the busiest international airport in the world, with passenger volume over double that of Gatwick.
Speaking to Computerworld UK at Splunk's .conf2017 in Washington D.C this week, Ibbitson explained how the challenges are "at a bigger scale at Dubai, not just for IT but across infrastructure".
The airport has also run out of room to physically expand beyond its current two-runway size. "Growth has to come through innovation and automation," Ibbitson said. "We don't want to be the biggest, we want to be the best, with new levels of customer service."
One of Ibbitson's first projects was to open an airport operations centre, bringing together all of the various operations teams like the airfield, transport, and IT. The idea is that everyone is "in one place" and sharing data across that facility with Splunk when it opens later this year.
The purpose-built facility will display real-time dashboards on large video screens and will have theatre-style seating.
Ibbitson explained: "We started a project to do the data integrations behind the scenes and start providing the meaningful data through applications or through a platform like Splunk, with one set of dashboards they can log into through web pages."
Driving customer experience levels
Dubai Airports has been using two deployments of Splunk Enterprise for a year now, one for its traditional IT operations and another to give more general operations analytic capabilities around queuing times and baggage flows.
The aim is to be able to monitor data across the whole estate - 4.5 billion data points and counting - so not just IT system logs but also data from sensors, air traffic control systems, transportation flows and other core infrastructure.
Ibbitson has set himself some ambitious goals. As well as reducing the airport's hefty energy consumption by 20 percent and offering the best airport wi-fi in the world, he's also aiming to cut security waiting times to below five minutes, much like the project he oversaw at Gatwick.
For energy consumption, Ibbitson has already started deploying sensors and using Splunk to get better visibility of energy usage so that his team can start to spot pinch points and identify areas for improvement. The goal is to cut its energy consumption - which is currently 2.5 percent of the whole city - by 20 percent, saving $25 million a year by 2023.
The first task is getting all of the data from the building management systems across the hundred buildings the airport operates. Splunk has given the airport a vehicle to consolidate that data and drive real-time information on energy to decision makers within the organisation so that they can start to consume energy more efficiently.
One project Ibbitson knows well is driving down waiting times at security. By ingesting data from ceiling sensors and 3D cameras, the airport can now see queue times in real time and start to blend historic data to make passenger volume predictions.
By giving this data to operations staff and allowing them to push notifications around spikes in passenger flows via its staff app, the airport can start to make the process more efficient and bring those queue times down.
Security staff at Dubai are not employed by the airport, they are police - so the airport decided not to provide analytics access to non-employees and instead gave dashboards to senior staff. This differed from the Gatwick approach where all security staff had access to passenger numbers via overhead monitors.
The airport has already seen wait times in transfer security halve since implementing Splunk and the new processes around it, down from averaging eight minutes to four.
Improving baggage handling
Now Ibbitson is really focusing on using data to build on the airport's baggage handling capacity, including by discovering and improving bottlenecks.
"We tied the operating system together and used Splunk to prove the changes we were making," he explained.
The airport now collects 200 data points per bag as each one travels the 15 kilometres of conveyor belts across the baggage system.
This gives staff the opportunity to spot issues that they didn't know existed before. For example, one section of the system was reporting timestamps for the year 2047, so the baggage system thought those bags had been in the system for a long time and started prioritising them unnecessarily.
According to Ibbitson, where the baggage system used to struggle with 10,000 bags an hour, the airport can now handle up to 13,000 an hour without any issues.
Further ahead, Ibbitson wants to start applying machine learning to this baggage data to predict where a bag is in the system 12-24 hours in advance, with the eventual aim being the capability to send an SMS to customers to say when and where their bag will land on the carousel.
Building the best airport wi-fi
After conducting some market research Dubai Airport identified that wi-fi was the number one concern of customers when it came to their overall satisfaction. "We wanted to have the best free airport wi-fi," Ibbitson said.
That would translate to free, streaming-quality wi-fi across the entire airport. By being able to monitor wi-fi usage and performance in real time, the IT operations team can start to spot rogue access points and identify black spots and target investment accordingly.
Finally, there's what the airport calls the "Golden Bathroom", a project to improve the cleanliness of the toilet facilities at the airport.
"To keep the toilets clean you have to understand how they are used," Ibbitson said. "There's no point cleaning a bathroom that isn't going to be used for three hours, you may as well save those cleaning operations for the hour when it is going to be most busy.
"So it was about finding where the busy peaks are and matching the cleaning contract to that."