McLaren Applied Technologies
Few companies are as highly respected as McLaren for their formidable capabilities in automotive engineering and racing. As with all modern sports, technology plays an important role in winning the game. McLaren saw an opportunity to apply its technology and insights to other fields, which was the genesis of McLaren Applied Technologies (MAT). Paul was hired as a consultant to assist with a key component of MAT’s growth strategy, called “Stream Processing.”
A vital element of McLaren’s racing ecosystem is the car-sensor network and data analytics used to measure racing performance in real time. It has become clear in recent years that we are on the cusp of a new wave of technological and business innovation related to large scale real-time sensing, which some refer to as the “The Internet of Things.” This is discussed at length in Paul’s new book “Connected Services,” where he also documents, in broad outline, the type of work that MAT is doing in this area, as described generally by their Managing Director, Geoff McGrath at Wired UK.
Due to confidentiality constraints, the project undertaken by Paul can’t be disclosed in any detail, but comes under the rubric of what some call “Complex Event Processing” and others call “Big Data” (also documented in the “Connected Services” book). Paul’s role was to provide external impetus to the product and architecture definition of an entirely new platform and computing paradigm, referred to as “Stream Processing.”
The “Internet of Things” (IoT) has three principle components: embedded low-cost real-time sensors, real-time communications, and hyper-scalable “Big Data” processing of the collected data (in the cloud). Paul is one of the pioneers in actually defining a commercial IoT product that is now in production.
This type of work marks a return for Paul to his roots in designing Digital Signal Processing (DSP) chips, which are essentially small-scale stream-processing machines. By exploiting the recent trends in low-cost cloud computing infrastructure, it is now possible to map the DSP idea to hyper-scale distributed machines where the ability exists to process millions, potentially billions, of real-time streams (as opposed to typically a few streams on a DSP chip).