Our research

© WiPLASH, EU Horizon 2020 project

Embedded systems are any applications that run on devices, from heart rate monitors to data centers. There are millions of ways to use embedded technology to improve our lives, and this is what drives the members of our lab!

David Atienza, Embedded Systems Lab

The long-term vision for ESL is to lead the future of intelligent connected systems by pioneering research on multi-objective co-design (HW-SW) of high-performance, energy-efficient, nano-scale edge AI architectures. Therefore, ESL covers full stack deployment of edge AI systems. A key component of our research is the deployment of open-source multi-processor system-on-chip (MPSoC) integrated circuits and platforms designed through the X-HEEP IC design framework and VersaSens platforms.

These are some of our flagship projects.

The main research lines within the ESL activities include, but are not restricted to, the following topics:

 

Circuits

Emerging memory technologies
Analog in Memory Cores
In and near-memory computing
in-SRAM computing
Software-defined SIMD
Near Memory Computing


Architectures and Systems

Accelerators
Coarse-Grained Reconfigurable Arrays
Hardware Accelerators for Posit Arithmetic
Analog in Memory Cores
Sustainable computing for SKAO
Systolic Array Accelerators for Transformers
Energy-efficient computing
Containergy
Federated machine learning over fog/edge/cloud architectures
Full system simulation and design
gXR5: A gem5-based full-system RISC-V simulator
Near-DRAM processors
WiPLASH: In-package Wireless Links
gem5-X: gem5 made easy
Cerberus: accuracy/run-time/thermal exploration of analog computing
Co-design for accelerated AI
Smart wearables
Inmodi knee brace
Bonsai: cough monitoring device
Epilepsy Monitoring
VersaSens
Systems-on-Chip
HEEPocrates, a tsmc65 implementation of X-HEEP
ML-enabled IoT devices and embedded AI
CHIMP
X-HEEP: an open-hardware RISC-V system-on-chip
Thermal modelling
Thermal-aware design of 2D/3D MPSoCs
Heterogeneous architectures for HPC applications
Fuel Cell Arrays: liquid cooling and power delivery
3D-ICE: 3D Interlayer Cooling Emulator


Algorithms

Cloud infrastructure management
Multi-objective management and optimization of servers and data centers
Edge artificial intelligence
HDTorch: HyperDimensional computing library
Edge Companions
Physiological data processing
DIGIPREDICT: multi-biosignals processing for cardiovascular conditions
Sparse sampling of bio-signals
Bonsai: cough monitoring device
Incremental inference and learning
Efficient federated learning


Applications

Cough Characterization
Bonsai: cough monitoring device
Coughvid
TB Diagnosis
Cough Benchmarks
Digital twins
DIGIPREDICT: multi-biosignals processing for cardiovascular conditions
UrbanTwin
Personalized health
Health and wellness monitoring
BiomedBench
Socio-Emotional and Ethical Blended Learning
Space applications
Embedded AI for aerospatial navigation
Sustainable computing for SKAO