The next generation of wireless networks, known as the fifth generation (5G), is a promising foundation for creating a reliable and robust infrastructure for a future digital society. 5G networks are based on key new technologies, such as massive multi-input multi-output (m-MiMo), that promise to increase network throughput and capacity and allow the network to cope with a large number of users.
m-MiMo, an extension of conventional MiMo, uses hundreds of antennas to reduce interference and prevent channel errors. The large number of antennas in m-MiMo also improves network security by enabling beamforming, a technique where the energy of the antennas is centralized in the direction of the legitimate receiver and an artificial noise (AN) signal is projected in the direction of malicious users.
Despite these benefits, there are still challenges that limit the applications of m-MiMo, including (1) channel contamination, (2) eavesdropping at the physical layer, and (3) excessive power consumption and complexity due to the use of a large number of elements of the radio frequency chain. Pilot sequence contamination is interference that results from processing the same pilot symbols in adjacent cells, whereas eavesdropping is an attack in which an intruding node intercepts messages exchanged between users, creating confidentiality problems in the network. The third constraint relates to energy mismanagement and to the processing complexity that is generated by elements of the radio frequency (RF) transmission chain.
The three constraints mentioned above are the subject of this thesis. Our main goal is to resolve issues related to channel contamination and physical layer security, as well as to optimize the number of antennas (parts of the radio frequency chain) in m-MiMo. To this end, we propose three models where we introduce Raptor codes, a form of error-correcting codes, into m-MiMo systems. The Raptor codes effectively reduce the magnitude of the problems associated with m-MiMo.
Raptor codes are a class of fountain codes that serve as error correctors in channel coding. These codes allow errors in a message to be detected and corrected without requiring the sender to retransmit the data, which is favorable in wireless networks where retransmission is not desirable due to channel errors.
To overcome the constraint of contaminated pilots, we first propose a model where the Raptor codes are combined with a minimum mean square error (MMSE) filtering technique to work with m-MiMo. In our model, the receiver uses the symbols decoded with the Raptor code to estimate the channel, rather than the pilot sequences sent by the transmitter. This solution helps estimate the channel while preventing pilot contamination. It also manages system energy more efficiently and reduces congestion in the network.
We tested our model using three forward error correction codes (Raptor, Luby transform, and low-density parity-check) and three filtering techniques (MMSE, zero forcing, and maximumratio combining). The study was performed on a slow fading channel, which either does not change over time or changes very little. Simulation results showed that our model achieves ideal performance using MMSE filtering with symbols decoded by the Raptor code.
We then propose a solution for preventing eavesdropping at the physical layer in m-MiMo. Our goal is to develop features that will allow Raptor codes to be used judiciously with m-MiMo for security purposes. Where appropriate, Raptor codes can be used as a tool by the intruder to reinforce his intrusion. Our proposal helps to secure the physical layer, avoid excessive power consumption, and reduce processing complexity. We tested our second model using a simulated intruder who had the most resources in terms of power, processing, and number of antennas, and who had perfect channel state information (CSI).
Our third chapter addresses the problem of excessive power consumption and processing complexity in m-MiMo by evaluating how to select the best subset of antennas at the receiver under imperfect CSI. We propose a water-filling algorithm based on a mutual information maximization criterion and on symbols decoded by Raptor code. Our method optimizes the number of antennas and, therefore, the number of frequency chain elements. Simulation results showed that our proposed solution achieves the same optimal values as the conventional exhaustive search method.
| Date | 3 Feb 2021 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Michel Kadoch (Supervisor) |
|---|
Benzid, D. (Author),
Kadoch (Supervisor),
3 Feb 2021Student thesis: Doctoral thesis › Doctorate in Engineering: Engineering