Wearable computing systems (WCS) are distributed systems of body-worn, computational, connected and sensor-based devices that collect physiological and behavioral data from users as well as contextual information from the environment. Thanks to their effectiveness and versatility, WCS have found diversified cyber-physical applications, including healthcare, assisted living, fitness, smart factories, and entertainment.
Former WCS (2000-2015) were based on wireless Body Sensor Networks (i.e. purposely designed programmable body sensors) with fixed or portable devices acting as base station; these solutions often resulted in resource-limited platforms, closed environments, operating stand-alone. More recently (>2013), with the diffusion of Cloud Computing, commercial wearable solutions have been implemented atop proprietary platforms: this allowed a gradual introduction in the market but hindering WCS from unleashing their full potential. Indeed, current WCS show several limitations: commercial platforms for wearable-enabled services are centralized, closed and often smartphone-dependent, thus resulting in vertical solutions with poor programmability, privacy and smartness. In fact, available Advanced RISC Machine / Microcontroller (ARM / MCU) based wearable devices do not have enough computing power to perform complex operations and WCS cannot prescind from Cloud Computing and its inherent drawbacks (privacy, responsiveness, etc).
To make WCS truly smart and Cloud/smartphone independent, COCOWEARS will define a framework for Computing Continuum WEARable Systems.
COCOWEARS relies on 3 pillars:
- sound theoretical foundations (including a hardware/software system co-design methodology based on simulations),
- tailored Artificial Intelligence/Machine Learning (AI/ML) algorithms,
- development tools aiming to enable the engineering of next-generation WCS.
The synergic exploitation of these pillars upon Field Programmable Gate Array devices (FPGA, due to their proven acceleration capabilities of AI/ML algorithms, despite their reasonable cost and dimensions makes COCOWEARS a holistic framework pursuing an innovative device-centric, yet device-independent, engineering approach. Indeed COCOWEARS promotes on-device intelligence at an unseen level but also enables a computing continuum, by pushing low-priority, more complex/global analytical tasks far from the user.
COCOWEARS represents the first solution fully exploiting AI technologies at different levels of granularity, complexity, and user proximity, able to push towards a massive adoption of WCS in the immediate future.