#	Pagina
attuale pagina	/open-h2020/projects/224058/results.html

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - DynapIP (Innovative and ultra-low power neuromorphic unit for advanced signal processing in wearable medical devices)

Teaser

Summary

Wearable devices are used to monitor numerous health parameters, however, the amount of data generated by wearables was 1500 petabytes (one petabyte = one million gigabytes) in 2018 and is expected to grow at a massive CAGR of 168% to reach 4020 petabytes in 2020 . Normally, this large amount of data needs to be processed externally using high-power cloud computation systems, implying high power consumption for both data transmission and processing, as well as large storage requirements. Commercially available health monitoring devices are inept at processing this data locally and are based on microprocessors commonly found in computers as central processing units (CPUs) or graphic processing units (GPUs), or on specialized microprocessors such as digital signal processors (DSPs). Transferring, storage and processing of this data is expensive, especially in continuous monitoring applications. Thus, a large chunk of the data is not processed and identification of pathological events is not achieved. Additionally, traditional processors imply high power consumption for data analysis and processing, hence it is not feasible to integrate them in wearables due to their lower battery capacities. DSPs are currently used to meet intensive on-board processing needs demanded by medical applications. However, DSPs imply a power consumption between 2 and 3.5W running at 1 GHz , limiting their use to devices with large batteries, limiting usable device lifetime between recharges, and making continuous processing highly energy expensive. During the phase 2 project, aiCTX will look to upgrade the input and output modules and configuration modules, as well as the software so different configurations can eventually be loaded on the DynapIP chip for real medical device use cases. A production line will be set up at CMO Global Foundries to produce 100 units for demonstration and validation with Biovotion a medical device company.

Work performed

aiCTX undertook a full analysis of both the technical and business potential of the DynapIP concluding that it would be highly valuable for the commercial industry. They defined necessary upgrades to the chip layout and software to maximize the efficiency of data processing within a wearable medical device. The company performed detailed analysis of the market size and different segments such as smart phones, autonomous vehicles etc. identified that the value of neuromorphic computing market is â‚¬2.03B in 2018 and is expected to reach â‚¬5.38Bn by 2023. â€¢ At a later stage, after initial penetration of the wearables devices market, aiCTX will look to diversify to other industries such as smartphones market, artificial intelligence, autonomous vehicle market and smart home devices market among others as DynapIP\'s technology is well suited for these emerging, high potential markets. aiCTX defined the commercialisation strategy and decided to commercialise initially to Europe specifically to Switzerland and France in the initial phase as the company has initial contact with companies in these countries followed by Germany and the UK. Both of these countries have a sizeable elderly population and also the largest number of wearable device users. Then, aiCTX will expand to Europe before expanding to other markets of North America and Asia.