Enhancing Underground Ad-Hoc Network Performance Under Dynamic Soil Conditions
We are looking for good candidates who are keen to improve wireless network performance in harsh and dynamic underground environement, and want to explore the use of machine learning in this domain.
The main research question is How to design an efficient, reliable, and adaptable wireless ad-hoc network routing algorithm that can cope with the conditions of soils, given low-frequency or LoRa modulation, humidity, and electric conductivity?
In fact, the dataset of the relationship between in-soil modulation schemes and the link budget created in previous experiments is available for this project, and can be used as a starting point. The main task is to design an algorithm to find the right routing path and modulation scheme based upon the soil condition to optimize the network performance using AI methods to make a model of the communication characteristics.
Problem Statement:
Wireless Underground Sensor Networks (WUSNs) usually target irrigation, earthquake monitoring, precision agriculture, intruder detection, assisted navigation, sport field maintenance, infrastructure monitoring, and environment monitoring applications. WUSN is a specialized kind of WSN that mainly focuses on the use of sensors buried sensors in the subsurface region of the soil. The main difference between WUSNs and the terrestrial WSNs is the communication medium. In particular, the differences between the propagation of electromagnetic (EM) waves in soil and in air are so significant that a complete characterization of the underground wireless channel was only available recently.
Despite its potential advantages, the realization of WUSN faces the main challenge which is the realization of efficient and reliable underground wireless communication between buried sensors, in an ad-hoc manner. Unlike stable terrestrial transmission, the soil's wireless underground communication links strongly depend on the soil condition. More specifically, the changes in temperature, weather, soil moisture, soil composition, and depth directly impact the connectivity and communication success in underground settings. Therefore the link budgets can vary according to the position and time. The right modulation scheme and a suitable communication path in a network can assure the quality of communication.
This project aims at developing an efficient underground ad hoc network scheme by using recurrent neural network to predict and control link quality given modulation schemes and soil conditions such as humidity and electric conductivity measured by sensors buried underground.
Requirements:
Machine learning and deep learning (basic is fine)
Arduino and python coding,
Knowing a network simulation tool is a plus
Work:
30% Theory, 50% Implementation and Experimental Validations, 20%Writing
Contact:
Assistant Professor Le Viet Duc (PS-EEMCS) (v.d.le@utwente.nl)
Daily supervisor: Jan Nguyen (PS-EEMCS) (d.t.n.nguyen@utwente.nl)
Office: Zilverling 5013