4 présentations

• 
  15h45 - 16h10

  Tensor-based Space Debris Parameter and Channel Estimation for Satellite Constellations

  • Adam Osmani, prés., Polytechnique Montréal
  • Gunes Karabulut Kurt, Polytechnique Montréal
  • Antoine Lesage-Landry, Polytechnique Montréal

  With the ever-growing need for efficient communication services, the number of satellites in orbit has drastically increased over the past decade. In order to implement these services in a sustainable way, it is essential to ensure that deployed satellites remain active for as long as possible. A notable obstacle to this goal is the increasingly high number of debris in orbit, possibly colliding or interfering with such satellites. In this work, we evaluate the performance of a tensor-based integrated communications and sensing space debris detection system. The algorithm allows the estimation of the angle of arrival (AoA), angle of departure (AoD), and time of arrival (ToA) of each incoming signal. The algorithm utilizes
  a CANDECOM/PARAFAC (CP) decomposition to extract a set of factor matrices from the received tensor. Each incoming signal is represented by a column vector in each matrix in the set. In the context of satellite communications these signals represent the different paths between the transmitter and receiver, such as the line-of-sight path and reflections caused by surrounding debris. A correlation-based scheme is then used to estimate the parameters associated with each path, which can then be used to determine the positions of nearby objects. To reduce the computational burden a sampling scheme is used, allowing for the manipulation of smaller tensors. The algorithm uses slab sampling to form the sampled tensors, corresponding to a partial use of available antennas accompanied by spaced intervals where all antennas are active. This scheme yields two tensors of different sizes, which can then be used to extract the desired parameters. Simulation results show that the root mean squared errors of extracted parameters increase with ambient noise radiance, and that sampling schemes sacrifice little to no accuracy.

• 
  16h10 - 16h35

  Connectivity-Aware Task Offloading for Remote Northern Regions: a Hybrid LEO-MEO Architecture

  • Mohammed Almekhlafi, prés., Department of Electrical Engineering, Polytechnique Montréal, QC, Canada
  • Antoine Lesage-Landry, Department of Electrical Engineering, Polytechnique Montréal, Mila & GERAD, QC, Canada
  • Gunes Karabulut Kurt, Department of Electrical Engineering, Polytechnique Montréal & Poly-Grames, QC, Canada

  Arctic regions, such as northern Canada, face significant challenges in achieving consistent connectivity and low-latency computing services due to the sparse coverage of Low Earth Orbit (LEO) satellites. To enhance service reliability in remote areas, this paper proposes a hybrid satellite architecture for task offloading that combines Medium Earth Orbit (MEO) and LEO satellites. We develop an optimization framework to maximize task offloading admission rate while balancing the energy consumption and delay requirements. Accounting for satellite visibility and limited computing resources, our approach integrates dynamic path selection with frequency and computational resource allocation. Because the formulated problem is NP-hard, we reformulate it into a mixed-integer convex form using disjunctive constraints and convex relaxation techniques, enabling efficient use of off-the-shelf optimization solvers. Simulation results show that, compared to a standalone LEO network, the proposed hybrid LEO-MEO architecture improves the task admission rate by 15% and reduces the average delay by 12%. These findings highlight the architecture’s potential to enhance connectivity and user experience in remote Arctic areas.

• 
  16h35 - 17h00

  Advanced Collision Avoidance and Trajectory Prediction for Drone Swarms in Urban Environments: Integrating Machine Learning and Optimization

  • Seyed Masoud Hashemi Ahmadi, prés., PhD candidate, Department of system Engineering, École de technologie supérieure (ÉTS)
  • Julio Montecinos, Professor, Department of system Engineering, École de technologie supérieure (ÉTS)

  Safe and efficient operation of autonomous drone swarms within complex urban airspaces poses significant challenges due to dynamic obstacles, unpredictable environmental conditions, and communication constraints. Traditional centralized collision avoidance methods often fail to provide adequate scalability, robustness, and real-time adaptability required for practical deployments. To address these gaps, this research proposes a novel decentralized framework integrating advanced machine learning and optimization techniques.
  Our methodological innovation involves synthesizing realistic drone trajectory data reflective of urban conditions, using it to train spatial-temporal predictive models combining Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN). These models anticipate the movements of both drone swarms and dynamic obstacles, thus facilitating timely collision avoidance decisions. We further propose novel optimization algorithms to enhance communication efficiency, enabling rapid and energy-conscious decision-making within decentralized drone networks.
  The anticipated outcomes of this research include improved predictive accuracy of drone trajectories, minimized collision risks, reduced computational latency, and optimized energy consumption. The methodological integration of machine learning models with optimization algorithms represents a significant advancement toward scalable, autonomous drone swarm operations, contributing directly to real-world applications in urban logistics, emergency response, and smart city traffic management.

• 
  17h00 - 17h25

  Optimization of Resilient Wireless Tactical Networks Using Tabu Search

  • Wissem Ahmed Zaid, prés., Polytechnique Montréal and GERAD
  • Alain Hertz, Polytechnique Montréal and GERAD

  This research focuses on optimizing wireless tactical networks using advanced heuristic methods, particularly Tabu Search. The objective is to design resilient communication infrastructures that ensure connectivity in environments where traditional telecommunications fail, such as disaster-stricken areas. These networks must adapt to unpredictable conditions while ensuring efficient data transmission and minimal latency.
  To tackle this problem, we adopt an integrated approach combining metaheuristics and graph theory-based techniques. The network topology is generated and optimized using Tabu Search, ensuring an adaptive and scalable structure. A geometric heuristic is then employed to configure antennas, maximizing signal strength and coverage while minimizing interference. Finally, the overall network configuration is refined through heuristics to enhance stability and efficiency.
  Results from synthetic tests demonstrate that our approach not only generates more effective network designs but also achieves this with a fast and efficient algorithm, significantly improving upon a previous solution. The study highlights the importance of heuristic-based approaches in addressing complex network design challenges and emphasizes the role of optimization techniques in ensuring efficient wireless communication, particularly in critical situations where connectivity is essential for emergency response and mission success.