Analysis and design of algorithms for the improvement of non-coherent massive MIMO based on DMPSK for beyond 5G systems
- Autores: Manuel José López Morales
- Directores de la Tesis: Ana García Armada (dir. tes.)
- Lectura: En la Universidad Carlos III de Madrid ( España ) en 2023
- Idioma: inglés
- Tribunal Calificador de la Tesis: Luis Castedo (presid.), Matilde Pilar Sánchez Fernández (secret.), Eva Lagunas Targarona (voc.)
- Programa de doctorado: Programa de Doctorado en Multimedia y Comunicaciones por la Universidad Carlos III de Madrid y la Universidad Rey Juan Carlos
- Materias:
- Enlaces
- Tesis en acceso abierto en: e-Archivo
- Resumen
Mobile communications have their earliest roots in 1947, when the Bell System business offered an early prototype of the current broadband mobile networks. Using very low-power transmitters, it was intended to deliver radio services across a wide area. Unfortunately, the Federal Communications Commission (FCC) refused to permit the use of the desired spectrum band. More than 30 years would pass before the first mobile communications system would be commercially available. The Bell Laboratories created the analog based Advanced Mobile Phone System (AMPS) based on frequency division multiple access (FDMA) during the 1980s, which was one of those pioneering technologies. The Global System for Mobile Communications (GSM) standard, the most popular 2G technology, was introduced later, in 1991. The analog voice service based on (FDMA) was replaced by the new digital service when digital time division multiple access (TDMA) emerged. With the launch of the GPRS and EDGE protocols, more functionality like SMS and other improvements were added over time. Nonetheless, the introduction of UMTS technology, or the 3G of mobile communications, marked the beginning of the multimedia revolution. There was a clear need to deliver more reliable and flexible communications while meeting significant data traffic needs, even though continuous advancements were made with the new mobile communications technology. In this context, LTE technology (almost-4G) began to be adopted quickly after it became standardized in the late 2000s. Orthogonal frequency division multiplexing (OFDM) and multiple-input-multiple-output (MIMO) are key technologies in LTE and its successive improvements carried in the second decade of the 21st century, which was gradually introducing new improvements until LTE-Advanced was obtained (true-4G). Since then, the technology continued evolving up until these days, in which the 5G New Radio standard was defined and started its deployment.
Wireless communications have gained more and more importance in our lives during the last 40 years. Nowadays, it is nearly impossible to think of a service that does not rely on them. Nowadays, the mobile networks are used by most of the society, and this is supported by the fact that by the end of 2021, about 4.3 billion people (more than half of the total population) actively use mobile internet. In fact, the mobile internet traffic represents about a 57% of the total global online traffic. Finally, there is an increasing trend both in the number of subscribers and in the traffic handled by each subscriber, caused by the fact that new services and applications emerge day after day and more and more people use them. The technology will keep evolving to allow for better and more demanding services and applications, which will motivate the development of even more capable technologies. The next foreseen mobile technology is beyond 5G currently under development with respect to the date of publication of this PhD thesis, followed by the 6G, which is expected to start its development around 2025. Mobile networks, along with numerous other telecommunications innovations, have connected every region of the planet. The world is one, as Arthur C. Clarke would remark.
A clear demand for larger data rates, smaller extreme-to-extreme delays and greater number of devices is a current issue for the development of mobile communications. Furthermore, it is foreseen that these demands should also be fulfilled in scenarios with stringent conditions, such as very fast varying wireless communications channels (either in time or frequency) or scenarios with power constraints, mainly found when the wireless equipment is battery powered.
Since most of the wireless communications techniques and standards rely on the fact that the wireless channel is somehow characterized or estimated to be pre/post-compensated in transmission and reception, there is a clear problem when the channels vary rapidly, or the available power is constrained. To estimate the wireless channel and obtain the so-called channel state information (CSI), some of the available resources (either in time, frequency, or any other dimension), are utilized by including known reference signals in the transmission and reception, typically called pilots, thus avoiding the use of these resources for data transmission. Another issue that limits the performance of the channel estimation is the pilot contamination problem, which consists of the fact that neighboring cells utilize the same resources to estimate the channel, interfering between them in the process.
If the channels vary rapidly, they must be estimated many times, which results in a very low data efficiency of the communications link. Also, in case the power is limited, the wireless link distance is large, or we have the problem of pilot contamination, the resulting signal-to-interference-and-noise ratio (SINR) will be low, which is a parameter that is directly related to the quality of the channel estimation and the performance of the data reception.
This problem is aggravated in massive multiple-input-multiple-output (MIMO), which is a promising technique for future wireless communications since it can increase the data rates, increase the reliability and cope with a larger number of simultaneous devices. In massive MIMO, the base-station (BS) is typically equipped with many antennas that are coordinated. In these scenarios, the channels must be estimated for each antenna, and thus, the problem of channel estimation aggravates. In time division duplex (TDD) this problem is simplified to at least each user due to channel reciprocity, but can still be problematic when there are many users. In this context, algorithms and techniques that can make the massive MIMO work without the use of CSI are of interest.
The proposed algorithms and techniques are based on the non-coherent massive MIMO, which allows to receive the data sent by a transmitter without the need to acquire the CSI. In this thesis, the non-coherent will be based on the use of DMPSK, by means of a differential detection of the received signals.
This thesis is focused on analyzing several properties of non-coherent massive MIMO and designing and implementing new algorithms to improve the performance of single user and multiuser non-coherent massive MIMO. All contributions can be summarized as follows: ¿ A cell-free (CF) approach is proposed for the single user non-coherent massive MIMO. The main idea is to make use of the access to several uncorrelated access points (APs) to increase the performance of the non-coherent massive MIMO. First, a detailed analysis of the effect of Rician spatially correlated channels on the performance of non-coherent massive MIMO is provided. Second, a set of techniques for AP selection in the proposed CF non-coherent massive MIMO are provided. Third, a comparison between coherent and non-coherent CF massive MIMO is done to demonstrate that the non-coherent approach is better in scenarios with stringent conditions.
¿ A blind channel estimation technique based on reconstructing the non-coherent data in reception to extract the channel information from the received signal is proposed. With this idea, first, an uplink hybrid scheme combining the non-coherent with the coherent scheme is proposed. In a classical pilot-symbol assisted modulation, the pilots are substituted by non-coherent data which will serve for both data transmission and channel estimation, thus increasing the efficiency of the uplink. Second, a pilot-less TDD massive MIMO is proposed, in which the uplink is only composed of non-coherent data with which the channel is estimated for precoding the downlink.
¿ Two constellation design techniques are proposed for multiuser non-coherent massive MIMO based on solving numerical optimization problems due to the intractability when trying to apply classical constellation design techniques. The first technique approximates the joint-constellation to a regular QAM shape and later conducts a bit mapping optimization over the individual constellations, while the second technique performs a Monte Carlo simulation of the performance allowing a joint-constellation and mapping design that does not make assumptions and is potentially optimal. A set of constellations is proposed with different number of users (up to 5), different constellation sizes (even between the users and up to 16-PSK) and different average received power.
