One of the key technologies that have emerged in recent years is the exploitation of unmanned aerial vehicles (UAVs) or drones in different applications ranging from the detection of hazardous materials in closed environments to the temporary deployment of flying base stations (also known as high-altitude platforms or HAPs) in future 5G and beyond networks. However, multiple challenges need to be addressed, including as one key example the accurate localization and synchronization of the UAVs. The problem becomes even more challenging in networks that deploy a large number of UAVs in swarm formations. Indeed, UAV swarms have to make their own decisions based on the information they share with each other. The aim of this project is to implement a fully synchronized swarm of UAVs or drones and to test its robustness to multiple scenarios requiring strict avoidance of obstacles indoor. The idea is to investigate new nature- or bio-inspired so-called “population-based” optimization techniques to ensure a fully functional swarm.

In this project, we aim to develop a new synchronization scheme for a swarm of drones or UAVs flying indoor in the presence of obstacles.
In the first phase of this project, we will build a peer-to-peer communication protocol that allows the relaying of multiple key parameters such as the RSSI and propagation delay. This type information will be fed to reinforcement learning technique that ensure the synchronization between the UAVs. A potential extension to this work is swarm synchronization in a heterogeneous setting – of great interest in Internet of things (IoT) applications – involving the cooperation between the swarm of UAVs or drones and ground vehicles.
In the second phase of this project, we will focus on assessing the performance of the newly proposed solutions under realistic scenarios. In this testing phase, we will use crazyflie drones in different swarm formations.
In the third and final phase of this project, we will evaluate the extension of the proposed synchronization scheme to the heterogeneous setting above. In this phase, we will combine the swarm of crazyflie drones with other moving units on the ground (e.g., rover robots).

Indoor localization is very often a daunting task, especially over non-line-of-sight (NLOS) transmission links. Many solutions have been so far proposed in the literature. Most promising ones rely on the so-called fingerprinting scheme combined with machine learning and, hence, require an offline process referred to as learning phase that needs to be reconducted each time the user moves from one environment to another. An interesting solution alternatively exploits the estimates of both the directions of arrivals (DOAs) and the propagation time delays (TDs) for localization. Such technique, however, is sensitive to the online estimation of a moving user.

In this project, we aim to develop a new online localization solution for moving users in indoor scenarios.
In the first phase of this project, we will investigate the improvement of Monte Carlo based joint DOAs and TDs estimation technique using a population based optimization solutions.
In the second phase of this project, we will focus on assessing the performance of the localization technique in a realistic environment. The DOAs and delays will later be fed to a localization technique to provide position estimates.
In the third phase of this project, we will focus on the assessment of the new localization technique in realistic environments that involve use of multiple unmanned aerial vehicles (UAVs). In this case, we will conduct two types of experiments. The first one uses Wi-Fi cards (e.g., Intel NIC and Atheros) mounted on a scientific-type drone (i.e., DJI Matrix). The second will exploit nano-drones (i.e., crazyflie) with their own “Loco Positioning” system.

Channel parameter estimation plays a crucial role in wireless communication systems. Parameters such carrier frequency and timing offsets can be exploited for synchronization purposes. Other parameters such as direction of arrival are well suited for localization. Most existing works very often rely on theoretical assumptions that still need to be tested using practical hardware equipment. One possible interesting solution is to exploit off-the-shelf Wi-Fi cards already integrated in today’s computers. These cards (e.g., Intel NIC and Atheros) require a specific driver to be installed in order to obtain the channel state information (CSI). However, such configuration could turn out to be impracticable for testing. Indeed, each of Wi-Fi cards mentioned above requires a full desktop which can a difficult task for evaluation of scenarios including mobile users.

In this project, we aim to use the a different type Atheros cards to create a network of multiple nodes for the testing of localization and channel parameter estimation techniques previously developed by our the team at the Wireless Lab.
In the first phase of this project, we aim to create a fully working network using different Atheros cards equipped on Arduino boards. This step includes the successful implementation of modified drivers, testing and monitoring the transmitted and received packets, and most importantly extracting the CSI using the modified drivers.
In the second phase of this project, we will focus on analyzing the extracted CSI at each receiving node. The CSI will be used to test compatible localization and channel parameter estimation techniques previously developed by our team.
In the third and final phase of this project, we will focus on adapting the Atheros card driver to combine the set Arduino boards with other moving units on the ground (e.g., rover robots that are also equipped with Arduino boards).