Dr. Shilong Lu and Fabrizio Lungo, researchers and developers at Fidelity Technologies Ltd., gave an informative presentation on Fidelity Technology Ltd.’s research strategies and product development, as well as detailing the challenges in developing an end-to-end real-time platform that provides meaningful and contextually appropriate interpretation of collected data, and delivers useful and actionable responses to the collected data.
Shilong Lu began the seminar by discussing the relevance of ‘Internet of Things’ (IoT), and ‘Machine-to-Machine’ (M2M) devices in emerging ‘Smart City’ development. Shilong outlined a few applications of the technologies such as light and temerature control in low-carbon buildings, or monitoring where personnel are so that non-critical systems in vacant areas can be deactivated to save energy. In healthcare, data collected from real time in vivo monitoring of patients could be sent to healthcare practitioners which could improve aftercare and long term patient monitoring. Analysis of the collected healthcare data could also be trained to warn of urgent health issues, or underlying conditions. Fidelity Technology approaches its solution from the point of view that customers don’t want to see the data, they want a meaningful interpretation or message from the the data they’re collecting.
Fabrizio Lungo then took over to discuss the current leaders in real-time software and hardware solutions. He discussed that the current packages deal only with either the data collection and transmission (hardware) or analysis of the data (software). Fabrizio then went on to explain that the limitation of the current solutions is that there is no complete end-to-end solution, which is what Fidelity Technology Ltd. is trying to achieve with their Plexus system. Fabrizio then gave an overview of the Plexus system architecture with details of the service that Plexus provides to its end users.
Fabrizio also gave an overview of the batch analysis frameworks Hadoop Map and Apache Spark, as well as the real-time analysis framework Apache Storm. He explained that batch processing is good if you have large data sets but that for continuous analysis of small blocks of real-time data, a real-time approach to the data analysis is needed as well.
Shilong and Fabrizio then opened up for a Q&A session. The first issue the NGCM members queried them on was the uptake of such real time systems in business from a retro-fitting and new equipment purchase point of view. They explained that both new equipment and retro-fitting were expensive options, and that until companies see the benefits of real-time data analysis they will be slow to invest in the market. The main source of concern from NGCM members was over security of the data streams and the application. They were assured that there were many security options in place to prevent malicious attacks on the system, the key security features involved security tunnels, only allowing access by recognised hardware, and only accepting recognised input patterns from the hardware. They also use encryption methods to further obfuscate the data and prevent unauthorised access to the devices in the network and the Plexus cloud itself.
Posted by Craig B. Rafter