Philips health cloud lead: ‘Privacy, compliance, upgradability shaping IoT architecture’

Ad Dijkhoff says the company's healthcare cloud ingests petabytes of data, experiencing 140 million device calls on its servers each data

Ad Dijkhoff says the company’s healthcare cloud ingests petabytes of data, experiencing 140 million device calls on its servers each day

Data privacy, compliance and upgradeability are having a deep impact on the architectures being developed for the Internet of Things, according to Ad Dijkhoff, platform manager HealthSuite Device Cloud, Philips.

Dijkhoff, who formerly helped manage the electronics giant’s infrastructure as the company’s datacentre programme manager, helped develop and now manages the company’s HealthSuite device cloud, which links over 7 million healthcare devices and sensors in conjunction with social media and electronic medical health record data to a range of backend data stores and front-end applications for disease prevention and social healthcare provision.

It collects all of the data for analysis and to help generate algorithms to improve the quality of the medical advice that can be generated from it; it also opens those datastores to developers, which can tap into the cloud service using purpose-built APIs.

“People transform from being consumers to being patients, and then back to being consumers. This is a tricky problem – because how do you deal with privacy? How do you deal with identity? How do you manage all of the service providers?” Dijkhoff said.

On the infrastructure side for its healthcare cloud service Philips is working with Rackspace and Alibaba’s cloud computing unit; it started in London and the company also has small instances deployed in Chicago, Houston and Hong Kong. It started with a private cloud, in part because the technologies used meant the most straightforward transition from its hosting provider at the time, and because it was the most feasible way to adapt the company’s existing security and data privacy policies.

“These devices are all different but they all share similar challenges. They all need to be identified and authenticated, first of all. Another issue is firmware downloadability – what we saw with consumer devices and what we’re seeing in professional spaces is that these devices with be updated a number of times during a lifetime, so you need that process to be cheap and easy.”

“Data collection is the most important service of them all. It’s around getting the behaviour of the device, or sensor behavior, or the blood pressure reading or heart rate reading into a back end, but doing it in a safe and secure way.”

Dijkhoff told BCN that these issues had a deep influence architecturally, and explained that it had to adopt a more modular approach to how it deployed each component so that it could drop in cloud services where feasible – or use on-premise alternatives where necessary.

“Having to deal with legislation in different countries on data collection, how it can be handled, stored and processed, had to be built into the architecture from the very beginning, which created some pretty big challenges, and it’s probably going to be a big challenge for others moving forward with their own IoT plans,” he said. “How do you create something architecturally modular enough for that? We effectively treat data like a post office treats letters, but sometimes the addresses change and we have to account for that quickly.”