Research Team

Project Leader

Photo of Boussaid Lotfi

Boussaid Lotfi

Position: Associate Professor

Involved Faculty Members

Haddaji Ramzi

Position: Assistant Professor (ENIM)

Bahri Nejmeddine

Position: Assistant Professor (FSM)

PhD Students to be mobilized within the project

Allami Waeel

Gaumati Yaseen

Nefzi Houimli Radhia

Hergli Seyfeddine

Hammouda Adel

Ben Amara Mohamed

Summary and Objectives

Preamble

The development of intelligent embedded systems for real-time applications is experiencing strong growth in response to growing needs at national and international levels, driven by sectors such as automotive, aeronautics, IoT, healthcare, Industry 4.0, precision agriculture, and meteorology.

It is essentially about developing optimized intelligent architectures on dedicated microcontrollers or FPGA prototyping circuits.

The major challenge of this part is to propose hardware solutions that increase processing parallelism and execution time, system flexibility and updatability, time-to-market reduction, and power consumption by adopting the AAA (Algorithm Architecture Adequacy) methodological approach which consists of optimizing the implementation of an algorithm on a specific hardware architecture.

Motivation

Intelligent embedded systems for real-time applications are evolving rapidly, driven by growing needs for local intelligence, connectivity, security, and energy efficiency.

Nationally and internationally, major trends include the integration of embedded artificial intelligence (Edge AI, TinyML) to enable local data processing, reducing latency and energy consumption while increasing device autonomy.

At the same time, cybersecurity is becoming a critical issue with the emergence of standards dedicated to the protection of sensitive embedded systems. Hardware architectures are evolving towards heterogeneous systems combining processors, GPUs, FPGAs, and AI accelerators to optimize performance for complex real-time applications.

Low-latency connectivity (notably via 5G and TSN networks) meets the needs of sectors such as autonomous vehicles, Industry 4.0, and connected healthcare. Finally, energy consumption reduction and eco-design have become priorities, particularly for IoT and mobile devices. These developments are accompanied by a strong need for research on mixed hardware/software design, real-time scheduling, operational safety, and critical system interoperability.

Objectives

Theme 1: Secure hardware accelerators for robotic communications

The first theme specifically targets the development of multi-level dedicated hardware accelerators to secure communications in robot control systems. Indeed, industrial robots and autonomous robots dedicated to real-time applications operating in critical environments, where purely software security solutions may prove insufficient, require the use of other hardware solutions that guarantee processing speed, computational accuracy, and security against various cyber threats.

Expected results include the development of functional hardware accelerators for secure communication modules, as well as a comparative performance analysis between existing hardware and software solutions in the field of communication security.

Theme 2: Intelligent photovoltaic supervision system

The second theme concerns the development of an intelligent supervision system for photovoltaic installations, aimed at optimizing energy production through Artificial Intelligence. This system will integrate fault prediction, identification, and localization functions to improve maintenance and overall plant performance. The project is structured around several axes: modeling of installations and faults for automated anomaly detection via AI, improvement of converters and inverters facing climatic variations through advanced MPPT algorithms, and intelligent optimization of energy storage based on consumption history and weather forecasts. Satellite image analysis will also enhance the prediction of solar production variations.

Theme 3: Medical decision support system

The third part of the research project aims to develop an intelligent medical decision support system based on artificial intelligence to improve the diagnosis of cardiovascular diseases in patients suffering from chest pain. The project combines in-depth analysis of ECG signals and pathophysiological data with the development of mortality predictive models using Artificial Intelligence technologies and language models. The final objective is to create a connected IoT bracelet prototype capable of collecting and analyzing cardiac signals in real time, predicting anomalies, and preventing the deterioration of patients' health status, thereby offering a continuous and personalized cardiovascular monitoring solution.

Theme 4: FPGA architecture for multimodal deep learning

The fourth theme aims to develop and optimize an FPGA-based architecture for executing multimodal deep learning models integrating audio, video, and text data. Faced with the complexity and significant computational resources required for these models, the research explores the advantages of FPGA circuits in terms of massive parallelism and specific configuration, using optimization techniques such as computational parallelism and inter-module communication optimization. The final objective is to demonstrate the superiority of this approach over traditional processors in terms of performance and energy efficiency, thus paving the way for a new generation of embedded systems capable of processing complex multimodal data in real time for applications in speech recognition, computer vision, and natural language processing.

Theme 5: Quantum computing on FPGA

The fifth and final part revolves around an in-depth study of the fundamental principles of quantum computing and key concepts related to FPGA circuits, allowing the examination of different possible architectures for a quantum computer based on this technology. The research particularly emphasizes the specific requirements of real-time applications, requiring optimal performance and responsiveness.

The major objective is to design and implement a functional prototype capable of efficiently executing quantum algorithms in real time, involving the design and optimization of logic circuits necessary to perform quantum operations, as well as the development of a user-friendly interface. This innovative approach opens promising perspectives for the future development of advanced computer systems capable of efficiently solving complex problems in various fields such as cryptography, optimization, and molecular simulation, thus contributing to the advancement of applied quantum technologies.

Summary

This project aims to design a new generation of intelligent embedded systems dedicated to real-time applications, integrating advanced hardware approaches and artificial intelligence. The project's themes generally focus on:

  1. The development of secure hardware accelerators for communications in robot control systems operating in critical environments, guaranteeing ultra-low latency and resistance to cyber threats.
  2. The creation of an AI system for photovoltaic supervision optimizing energy production through fault prediction, converter improvement (advanced MPPT), and intelligent storage management.
  3. The design of a connected medical bracelet for real-time cardiovascular diagnosis, analyzing ECG signals and physiological data via AI predictive models.
  4. The optimization of FPGA architectures for accelerated execution of multimodal models (audio/video/text), aiming for energy efficiency superior to conventional processors.
  5. The exploration of quantum architectures on FPGA for real-time computing, including a prototype executing quantum algorithms with a dedicated user interface.

The whole is structured around a synergy between reconfigurable hardware (FPGA, ASIC circuits) and AI, aiming to push the limits of embedded systems in critical domains: robotics, renewable energy, healthcare, multimodal, and quantum computing. Expected results include functional prototypes validating gains in performance, security, and energy consumption compared to traditional software solutions.

Research Program and Methodology

Methodological Approach

The new circuits to be developed will be tested on dedicated FPGA-based prototyping modules by adopting the Algorithm Architecture Adequacy (AAA) approach and using meteorological data provided by INM (National Institute of Meteorology) and others acquired from the weather station acquired within the framework of the FALSAFA project.

Monthly reports and presentations will be presented to monitor progress, scientific production, and difficulties encountered by doctoral students.

Research internships in foreign laboratories could be scheduled to strengthen collaboration and ensure better scientific production.

Project Implementation Timeline

First Year

Literature Review & Definition of Objectives

Literature review step:

  • Conduct an in-depth review of existing literature on the studied theme.
  • Identify the most relevant research work in the field and understand current advances.
  • Comparative study of the most relevant approaches from an A3 perspective.

Definition of objectives step:

  • Clearly define the specific objectives of the work to be carried out such as design, performance optimization particularly for real-time applications, etc.
Second Year

Design & Simulation

Design step:

  • Decompose the algorithm(s) to be implemented in a context of intensive parallelism.
  • Design the architecture of prototypes and models optimally.

Simulation and verification step:

  • Use simulation tools to verify the proper functioning of prototypes and models.
  • Perform exhaustive tests to validate the obtained results.
Third Year

Optimization & Implementation

Performance optimization step:

  • Analyze prototype and model performance in terms of speed, accuracy, energy consumption, etc.
  • Identify weak points and propose improvements to optimize these performances.

Hardware implementation step:

  • Transform the prototype design into a real hardware implementation using appropriate electronic components.
  • Conduct experimental tests based on this hardware implementation.

Experimental evaluation step:

  • Carry out experiments to evaluate the performance of the realized prototypes.
  • Compare these performances with those obtained by other existing approaches.
Fourth Year

Analysis & Finalization

Results analysis and thesis writing step:

  • Analyze the obtained experimental results and draw relevant conclusions.
  • Write the thesis following appropriate academic standards, including an introduction, a state of the art, a detailed description of the work carried out, a critical analysis of the results, a conclusion, and perspectives.

Finalization:

  • Finalization of prototypes
  • Overall analysis of project impacts
  • Quantify results in terms of theses, scientific papers, patents, etc.
  • Inventory and analysis of unfinished points
  • Summary report.

Cooperation and Partnership

Cooperation with foreign laboratories

  • Hosting researchers for internships.
  • Co-supervision of PhD students.
  • Joint publications and joint communications.

Expected Results (publications, patents, theses, habilitations, ...)

Scientific Publications

In terms of scientific publications, the studied themes could be the source of several articles in well-known scientific journals in the field.

Theme 1

Articles on cyber-secure hardware architectures for robots (latency/attack resistance benchmarks).

Theme 2

Publications in AI for renewable energy (predictive photovoltaic fault models + MPPT/storage optimization).

Theme 3

Papers in medical journals (The Lancet Digital Health) on real-time cardiac diagnostic algorithms.

Theme 4

Research on FPGA optimization for multimodal models (neural network compression, parallelism).

Theme 5

Pioneering studies in Quantum Information Processing on quantum emulation via FPGA.

Patents

Regarding potential patents:

  • Security hardware accelerators (theme 1): Encryption architecture dedicated to robotic communications.
  • AI algorithm for photovoltaic fault prediction (theme 2) + Adaptive energy storage system.
  • Connected medical bracelet (theme 3): Cardiac risk prediction method via ECG + AI.
  • FPGA architecture for multimodal processing (theme 4): Optimization of audio/video/text data flows.
  • Quantum computer on FPGA (theme 5): Logic circuit design for emulated quantum gates.

Prototypes and Demonstrators

  • Security hardware accelerators (theme 1): Encryption architecture dedicated to robotic communications.
  • AI algorithm for photovoltaic fault prediction (theme 2) + Adaptive energy storage system.
  • Connected medical bracelet (theme 3): Cardiac risk prediction method via ECG + AI.
  • FPGA architecture for multimodal processing (theme 4): Optimization of audio/video/text data flows.
  • Quantum computer on FPGA (theme 5): Logic circuit design for emulated quantum gates.

Career Advancement

  • Opening of a professor position
  • Opening of two associate professor positions

Socio-economic Benefits of the Project

Industry 4.0

Enhanced Cybersecurity: Hardware accelerators for critical robots (theme 1) will reduce cyberattack risks in smart factories, protecting production lines and preventing economic losses.

Energy Efficiency: Optimization of embedded systems (themes 4-5) will decrease factory energy consumption via low-carbon footprint FPGA/quantum controllers.

Industrial Competitiveness: Secure and responsive robots will enable more flexible production, adapted to volatile markets (e.g., rapid reconfiguration of assembly lines).

Precision Agriculture

Smart Crop Monitoring: The multimodal FPGA architecture (theme 4) will enable real-time analysis of drone data (images, soil sensors) to detect diseases/water stress, reducing pesticide use (+20% efficiency according to FAO studies).

Resource Optimization: Advanced MPPT algorithms (theme 2) will improve the energy supply of autonomous solar farms, while predictive AI will adjust irrigation based on weather data.

Increased Profitability: Reduction of input costs (water, fertilizers) and decrease in crop losses (-15% according to the World Economic Forum).

Meteorology

Hyperlocal Forecasts: FPGA quantum computers (theme 5) will accelerate the processing of complex climate models, enabling early warnings for extreme events (floods, droughts).

Optimized Satellite Data: AI analysis of satellite images (theme 2) will refine solar production forecasts and microclimatic impact on territories.

Food Security: Reliable weather will support agricultural planning in vulnerable regions (e.g., sub-Saharan Africa).

Sustainable Energy Production

Renewable Energy Boost: The AI photovoltaic supervisor (theme 2) will increase plant yield (+10-15%) through predictive maintenance and optimized storage management, accelerating the energy transition.

Resilient Electrical Grids: The integration of secure embedded systems (theme 1) will protect critical infrastructures against cyber threats.

Energy Access: AI-optimized solar microgrids will facilitate rural electrification (potential: +300 million connected people by 2030, World Bank).

Cross-cutting Impacts (sustainable development)

Job Creation: more than 5,000 skilled positions in green tech (robotics, energy, digital health) by 2030.

CO2 Emission Reduction: -2.5 Mt/year through embedded system energy efficiency and massive deployment of optimized solar energy.

Technological Sovereignty: Reduction of dependence on foreign chips through local FPGA/quantum architectures.

Social Inclusion: The medical bracelet (theme 3) will democratize access to cardiac care in isolated areas.

Synthesis

This project positions embedded innovation (AI + hardware) as a lever for decarbonized, resilient, and inclusive growth, increasing access to decent jobs (SDG 8). Direct economic gains (cost reduction and new markets) will be remarkable according to studies on similar work carried out by the European Commission on green digitalization.