ENTRUST
usEr ceNtric plaTform foR continoUS healThcare

Exploiting Edge Intelligence for the Health Internet of Things

Vision

Edge Intelligence for continous health monitoring

The research project means to exploit both architectural and algorithmic challenges related to the development and deployment of edge-based Health Internet of Things(HIoT) systems. We propose ENTRUST Platform with a fluid edge-to-cloud architecture purposely tailored for addressing the complexity of HIoT and related applications in terms of requirements and goals (from responsiveness and usability to privacy and interoperability).

Specific Objectives

Implementing edge-to-cloud-continuum in Health IoT Systems

Interoperability and integrability

The joint exploitation of gateway and software adapters designed in a modular fashion allows supporting different communication protocols and technologies.

Efficiency and effectiveness

Driven by the simulation, the deployment becomes a systematic task, no longer based upon empirical considerations but on clear indications.

AI optimization techniques for HIoT

For HIoT purposes, model compression techniques are being exploited. Our simulation driven approach allows considering different target hardware with different degrees of parallelism in order to avoid unnecessary model pruning that could lead to lower performance and accuracy.

Challenges

Towards the development of ENTRUST Platform

General challenges in HIoT concern many engineering aspects

Wearables and innovative approaches to data fusion

Edge computing for data pre-processing, filtering and compression

Stream processing and parallel architecture

Cloud computing for intensive processing/data storage

Machine learning for anomaly detection

Specific challenges

Simulation complexity

Edge-cloud continuum with fine grained and context-aware offloading policies, support for mobility and energy features, models for the mainstream communication protocols.

Continous monitoring

Typical AAL and IoMT systems work in “controlled” environments. ENTRUST exploits all the resources within the user availability so that anytime the user would find the better support for his/her service execution.

Edge infrastructure for early detection of anomalies

Signals from wearable sensors and smart clothes can be processed at a wearable or portable level in the local layer by the use of small custom microcontrollers or smartwatches to allow data filtering and AI analysis in mobility.

Planning

Milestones

Analysis

This milestone focuses on the literature review and on requirements analysis aiming at the identification of key building blocks for the ENTRUST platform design.

Design

This milestone concerns all activities related to the release of a first version for each component of the ENTRUST platform: the architecture definition, the simulator customization , the AI Models definition.

Implementation

This milestone focuses on the finalization of the ENTRUST platform and particularly on the implementation, validation and deployment of the AI model.

People

Our team

Claudio Savaglio

Associated Principal Investigator

Assistant Professor
DIMES Department
University of Calabria

Giuseppe D'Aniello

UNISA Research member

Assistant Professor
DIEM Department
University of Salerno

Francesco Pupo

UNICAL Research Member

Associate Professor
DIMES Department
University of Calabria

Consortium

Research units

NEWS

Latest posts

Recruiting @UNICAL!

Read More

Kickoff meeting!

Our contribution @DISCOLI workshop // DCOSS-IoT 2024

Read More