In the sixth generation of wireless communication systems (6G), there exist multiple candidate enabling technologies that help the wireless network satisfy the ever-increasing demand for speed, coverage, reliability, and mobility. Among these technologies, reconfigurable intelligent surfaces (RISs) extend the coverage of a wireless network into dead zones, increase capacity, and facilitate integrated sensing and communications (ISAC) tasks by consuming very low power, thus contributing to energy efficiency as well.

RISs are meta-material-based devices whose electromagnetic reflection characteristics can be controlled externally to cater to the needs of the communication links. Most ubiquitously, this comes in the form of adding a desired phase shift to an incident wave before reflecting it, which can be used to phase-align multiple incident waves to increase the strength of the signal at the receiver and provide coverage to an area that otherwise would be a dead zone.

While this portrays an image of a dream technology that would boost the existing wireless networks significantly, RISs do not come without engineering problems. First of all, the individual elements do not exhibit ideal reflection characteristics, that is, they attenuate the incident signal in a fashion depending on the configured phase shift. This creates the phenomenon called "phase-dependent amplitude". Another problem caused by RISs is the channel estimation overhead. In a multiple-antenna communication system, the channel between two terminals is as complex as the product of the number of antennas at each end. However, when an RIS comes into the equation, the cascade of the transmitter-RIS and RIS-receiver channels has a complexity further multiplied by the number of RIS elements. Consequently, the channel estimation process to utilize the RIS effectively becomes more demanding, that is, more pilot signals are required to estimate the channel for coherent reception. This adversely affects the effective data rate within a communication system since more resources need to be spent for pilot transmission and fewer resources can be allocated for data transmission. While there exists some work on reducing the channel dimensions by exploiting the channel structure, this problem persists for unstructured channels. In addition, for the wireless networks using multiple RISs, a new kind of pilot contamination arises, which is the main topic of this thesis.

In the first part of this thesis, we study this new kind of pilot contamination in a multi-operator context, where two operators provide services to their respective served users and share a single site. Each operator has a single dedicated RIS and they use disjoint frequency bands, but each RIS inadvertently reflects the transmitted uplink signals of the user equipment devices in multiple bands. Consequently, the concurrent reflection of pilot signals during the channel estimation phase introduces a new inter-operator pilot contamination effect. We investigate the implications of this effect in systems with either deterministic or correlated Rayleigh fading channels, specifically focusing on its impact on channel estimation quality, signal equalization, and channel capacity. The numerical results demonstrate the substantial degradation in system performance caused by this phenomenon and highlight the pressing need to address inter-operator pilot contamination in multi-operator RIS deployments. To combat the negative effect of this new type of pilot contamination, we propose to use orthogonal RIS configurations during uplink pilot transmission, which can mitigate or eliminate the negative effect of inter-operator pilot contamination at the expense of some inter-operator information exchange and orchestration.

In the second part of this thesis, we revisit the inter-operator pilot contamination. This time, however, we investigate the use of multiple antennas at the base stations to eliminate inter-operator pilot contamination. While orthogonalizing the RIS configurations as in the single antenna case eliminates pilot contamination, it doubles the number of pilots required to perform channel estimation. Considering the extant pilot overhead problem in RIS-aided communication systems, an alternative approach that does not increase the number of pilots to be transmitted is necessary. To this end, we propose using receive beamforming and null forming to eliminate inter-operator pilot contamination. We show that it is possible to eliminate inter-operator pilot contamination by placing nulls toward the signal coming from the other operator's RIS.

In the third part of this thesis, we consider a single-operator-two-RIS ISAC%integrated sensing and communication (ISAC) system where the single user is both a communication terminal and a positioning target. Based on the uplink positioning pilots, the base station aims to estimate both the communication channel and the user's position within the indoor environment by estimating the angle of arrival (AoA) of the impinging signals on both RISs and then exploiting the system and array geometries to estimate the user position and user channels respectively. Although there is a single operator, due to the presence of multiple RISs, pilot contamination occurs through the same physical means as multi-operator pilot contamination unless the channel estimation process is parameterized. Since the communication links are considered to be pure line-of-sight (LOS), their structure allows the reduction of the number of unknown parameters. Consequently, the reduction of information caused by pilot contamination does not affect the channel estimation procedure, hence the pilot contamination is overcome. In addition, the position of the user is determined by intersecting the lines drawn along the AoA estimates. We adopt the Cramér-Rao Lower Bound (CRLB), the lower bound on the mean squared error (MSE) of any unbiased estimator, for both channel estimation and positioning. Our numerical results show that it is possible to utilize positioning pilots for parametric channel estimation when the wireless links are LOS.

The fourth part of the thesis ventures into the domain of near-field communications. Here, we consider the estimation of parametric channels in the uplink of a multi-user multiple-input-multiple-output (MU-MIMO) communication system where the users are located within the radiative near field (Fresnel region) of the base station's aperture antenna. In this setup, we consider near-field channel models characterized by the users' distances and azimuth angles relative to the aperture array. We derive the CRLB to estimate these location parameters and the parametric channel estimates in closed form. Moreover, we consider using the 2D-MUSIC algorithm to estimate these parameters and compare the performance of the 2D-MUSIC algorithm with the CRLB. Our results indicate that the 2D-MUSIC algorithm is asymptotically consistent and efficient.

I den sjätte generationen av trådlösa kommunikationssystem (6G) finns flera potentiella tekniker som gör det möjligt för det trådlösa nätverket att uppfylla de ständigt ökande kraven på hastighet, täckning, tillförlitlighet och rörlighet. Bland dessa tekniker återfinns de så kallade omkonfigurerbara intelligenta ytor (reconfigurable intelligent surface - RIS på Engelska), vilka kan förlänga ett nätverks täckning till områden utan signal (så kallade ”dead zones”), öka kapaciteten och underlätta integrerade sensor- och kommunikationslösningar (ISAC). Samtidigt förbrukar RIS mycket lite energi och bidrar därmed till ökad energieffektivitet.

RIS är en metaytebaserad teknologi där ytan kan styras externt för att justera hur inkommande elektromagnetiska vågor reflekteras. Vanligtvis innebär detta att varje element på ytan åstadkommer en önskad fasförskjutning av den inkommande signalen innan den reflekteras, vilket kan utnyttjas för att fassjustera flera inkommande vågor och därmed förstärka signalen vid mottagaren. På så sätt kan täckning ges till annars otillgängliga områden.

Trots att denna teknik lovar betydande förbättringar för dagens trådlösa nätverk medför RIS även nya ingenjörsmässiga utmaningar. För det första uppvisar de enskilda elementen inte perfekta reflektionskarakteristiker, utan dämpar signalen på ett sätt som beror på den inställda fasförskjutningen, ett fenomen som kallas fasberoende amplitud. För det andra uppstår extra overhead inom kanalskattningen. I ett flerantennsystem är kanalens komplexitet mellan två noder redan omfattande, men när en RIS läggs till multipliceras komplexiteten med antalet RIS-element. Följaktligen krävs fler pilotsignaler för att skatta kanalen i system som använder RIS, vilket minskar den effektiva datahastigheten eftersom fler resursblock går åt till pilotsändning. Även om det finns metoder för att reducera kanalens dimensioner genom att utnyttja kanalstruktur, kvarstår problemet för kanaler utan särskild struktur. Därtill uppstår en ny typ av pilotförorening (pilot contamination) i scenarier med flera RIS:ar eller flera operatörer, vilket är huvudtemat i denna avhandling.

I avhandlingens första del studeras denna nya typ av pilotförorening i ett multioperatörsscenario, där två operatörer delar samma plats men använder olika frekvensband och har varsin dedikerad RIS. Även om frekvensbanden är åtskilda reflekterar varje RIS oavsiktligt de upplänksignaler som skickas av användare i båda banden. Under kanalskattningsfasen leder denna samtidiga reflektion av pilotsignaler till interoperatörs-pilotförorening. Vi analyserar hur fenomenet påverkar system med både deterministiska och korrelerade Rayleigh-fadande kanaler, med fokus på kanaluppskattning, signalequalisering och kapacitet. Numeriska resultat visar en tydlig prestationsförsämring och understryker att interoperatörs-pilotförorening är ett allvarligt problem i multioperatörssystem med RIS. För att motverka denna effekt föreslår vi att använda ortogonala RIS-konfigurationer under upplänkens pilotfas. Detta kan mildra eller helt eliminera pilotföroreningen, men kräver samordning och viss informationsdelning mellan operatörerna.

I avhandlingens andra del studeras samma interoperatörs-pilotförorening men i ett scenario där basstationerna har flera antenner. Att ortogonalisera RIS-konfigurationerna, som i fallet med en enda antenn, eliminerar visserligen pilotföroreningen, men fördubblar samtidigt antalet nödvändiga pilotsignaler. Mot bakgrund av den redan existerande pilotsignalsproblematiken i RIS-baserade system behövs en annan lösning som inte ökar pilotbehovet. Därför undersöks här hur mottagarstrålformning och nollställning (null forming) kan användas för att eliminera pilotföroreningen. Vi visar att det är möjligt att undertrycka den oönskade signalen från den andra operatörens RIS genom att placera nulls mot den, och därmed undvika pilotförorening utan att öka antalet pilotsignaler.

I avhandlingens tredje del betraktas ett enoperatörssystem med två RIS:ar i en integrerad sensor- och kommunikationsmiljö (ISAC), där en enda användare fungerar både som kommunikationsenhet och positioneringsmål. Baserat på upplänkens positioneringspiloter försöker basstationen uppskatta både kommunikationskanalen och användarens position inomhus. Detta görs genom att först skatta infallsvinklar (AoA) på signalerna som träffar de två RIS:arna och därefter, med hjälp av system- och antenngeometrier, beräkna användar-positionen samt motsvarande kanaler. Trots att detta inte är ett multioperatörsscenario kan pilotförorening uppstå av samma fysiska skäl när flera RIS:ar används, såvida kanaluppskattningen inte parameteriseras. Eftersom kommunikationslänkarna här antas vara rena siktlänkar (LOS) kan man dock utnyttja kanalstrukturen för att minska antalet okända parametrar. Detta gör att informationsförlusten orsakad av pilotförorening inte påverkar kanaluppskattningen, och pilotföroreningen kan därmed överbryggas. Vidare bestäms användarens position genom att skära de räta linjerna som definieras av AoA-estimaten. Vi använder Cramér-Raos lägsta gräns (CRLB) som prestandamått för den lägsta möjliga medelkvadratfelet (MSE) för en %obiaserad skattning av både kanal och position. Numeriska resultat visar att det är fullt möjligt att utnyttja positioneringspiloter för parameterbaserad kanaluppskattning i LOS-scenarier.

I avhandlingens fjärde del förflyttar vi oss till närfältskommunikation, där användarna befinner sig i den s.k. Fresnelzonen relativt basstationens antennapertur. Vi analyserar ett uplänkscenario i ett fleranvändarsystem med flera antenner (MU-MIMO) och antar att kanalmodellen är närfältsbaserad och beror på användarnas avstånd och azimutvinkel relativt antennaperturen. Vi härleder CRLB för att i slutet uppskatta såväl positioneringsparametrar (avstånd och vinkel) som de parameteriserade kommunikationskanalerna i slutet form. Dessutom undersöker vi hur 2D-MUSIC-algoritmen kan användas för att skatta dessa parametrar, och vi jämför dess prestanda med CRLB. Resultaten visar att 2D-MUSIC är både asymptotiskt konsistent och effektiv för denna typ av närfältsbaserad kanaluppskattning.

In the sixth generation of wireless communication systems (6G), there exist multiple candidate enabling technologies that help the wireless network satisfy the ever-increasing demand for speed, coverage, reliability, and mobility. Among these technologies, reconfigurable intelligent surfaces (RISs) extend the coverage of a wireless network into dead zones, increase capacity, and facilitate integrated sensing and communications tasks by consuming very low power, thus contributing to energy efficiency as well.

RISs are meta-material-based devices whose electromagnetic reflection characteristics can be controlled externally to cater to the needs of the communication links. Most ubiquitously, this comes in the form of adding a desired phase shift to an incident wave before reflecting it, which can be used to phase-align multiple incident waves to increase the strength of the signal at the receiver and provide coverage to an area that otherwise would be a dead zone.

While this portrays an image of a dream technology that would boost the existing wireless networks significantly, RISs do not come without engineering problems. First of all, the individual elements do not exhibit ideal reflection characteristics, that is, they attenuate the incident signal in a fashion depending on the configured phase shift. This creates the phenomenon called "phase-dependent amplitude". Another problem caused by RISs is the channel estimation overhead. In a multiple-antenna communication system, the channel between two terminals is as complex as the product of the number of antennas at each end. However, when an RIS comes into the equation, the cascade of the transmitter-RIS and RIS-receiver channels has a complexity further multiplied by the number of RIS elements. Consequently, the channel estimation process to utilize the RIS effectively becomes more demanding, that is, more pilot signals are required to estimate the channel for coherent reception. This adversely affects the effective data rate within a communication system since more resources need to be spent for pilot transmission and fewer resources can be allocated for data transmission. While there exists some work on reducing the channel dimensions by exploiting the channel structure, this problem persists for unstructured channels. In addition, for the wireless networks using multiple RISs, a new kind of pilot contamination arises, which is the main topic of this thesis.

In the first part of this thesis, we study this new kind of pilot contamination in a multi-operator context, where two operators provide services to their respective served users and share a single site. Each operator has a single dedicated RIS and they use disjoint frequency bands, but each RIS inadvertently reflects the transmitted uplink signals of the user equipment devices in multiple bands. Consequently, the concurrent reflection of pilot signals during the channel estimation phase introduces a new inter-operator pilot contamination effect. We investigate the implications of this effect in systems with either deterministic or correlated Rayleigh fading channels, specifically focusing on its impact on channel estimation quality, signal equalization, and channel capacity. The numerical results demonstrate the substantial degradation in system performance caused by this phenomenon and highlight the pressing need to address inter-operator pilot contamination in multi-operator RIS deployments. To combat the negative effect of this new type of pilot contamination, we propose to use orthogonal RIS configurations during uplink pilot transmission, which can mitigate or eliminate the negative effect of inter-operator pilot contamination at the expense of some inter-operator information exchange and orchestration.

In the second part of this thesis, we consider a single-operator-two-RIS integrated sensing and communication (ISAC) system where the single user is both a communication terminal and a positioning target. Based on the uplink positioning pilots, the base station aims to estimate both the communication channel and the user's position within the indoor environment by estimating the angle of arrival (AoA) of the impinging signals on both RISs and then exploiting the system and array geometries to estimate the user position and user channels respectively. Although there is a single operator, due to the presence of multiple RISs, pilot contamination occurs through the same physical means as multi-operator pilot contamination unless the channel estimation process is parameterized. Since the communication links are considered to be pure line-of-sight (LOS), their structure allows the reduction of the number of unknown parameters. Consequently, the reduction of information caused by pilot contamination does not affect the channel estimation procedure, hence the pilot contamination is overcome. On the other hand, the position of the user is determined by intersecting the lines drawn along the AoA estimates. We adopt the Cramér-Rao Lower Bound (CRLB), the lower bound on the mean squared error (MSE) of any unbiased estimator, for both channel estimation and positioning. Our numerical results show that it is possible to utilize positioning pilots for parametric channel estimation when the wireless links are LOS.

In this paper, we study a new kind of pilot contamination appearing in multi-operator reconfigurable intelligent surfaces (RIS) assisted networks, where multiple operators provide services to their respective served users. The operators use dedicated frequency bands, but each RIS inadvertently reflects the transmitted uplink signals of the user equipment devices in multiple bands. Consequently, the concurrent reflection of pilot signals during the channel estimation phase introduces a new inter-operator pilot contamination effect. We investigate the implications of this effect in systems with either deterministic or correlated Rayleigh fading channels, specifically focusing on its impact on channel estimation quality, signal equalization, and channel capacity. The numerical results demonstrate the substantial degradation in system performance caused by this phenomenon and highlight the pressing need to address inter-operator pilot contamination in multi-operator RIS deployments. To combat the negative effect of this new type of pilot contamination, we propose to use orthogonal RIS configurations during uplink pilot transmission, which can mitigate or eliminate the negative effect of inter-operator pilot contamination at the expense of some inter-operator information exchange and orchestration.

In this letter, we investigate the use of reconfigurable intelligent surfaces (RISs) to jointly estimate the position and channel of a user equipment (UE) using uplink pilot signals. We consider a setup with a user and a base station (BS), where the direct path between the BS and the UE is blocked and virtual line-of-sight (LOS) links are created over two reconfigurable intelligent surfaces (RISs). We investigate the benefits of exploiting the channel geometry to estimate the user’s position and the user-RIS channels jointly in terms of estimation performance and pilot overhead. To this end, we consider the Cramér-Rao Lower Bound for channel estimation and UE localization. Our numerical results show that exploiting the LOS structure of the channels improves the channel estimation performance by several orders of magnitude and reduces the channel estimation performance by reducing the number of unknown parameters.

We introduce a novel deep reinforcement learning (DRL) approach to jointly optimize transmit beamforming and reconfigurable intelligent surface (RIS) phase shifts in a multiuser multiple input single output (MU-MISO) system to maximize the sum downlink rate under the phase-dependent reflection amplitude model. Our approach addresses the challenge of imperfect channel state information (CSI) and hardware impairments by considering a practical RIS amplitude model. We compare the performance of our approach against a vanilla DRL agent in two scenarios: perfect CSI and phase-dependent RIS amplitudes, and mismatched CSI and ideal RIS reflections. The results demonstrate that the proposed framework significantly outperforms the vanilla DRL agent under mismatch and approaches the golden standard. Our contributions include modifications to the DRL approach to address the joint design of transmit beamforming and phase shifts and the phase-dependent amplitude model. To the best of our knowledge, our method is the first DRL-based approach for the phase-dependent reflection amplitude model in RIS-aided MU-MISO systems. Our findings in this study highlight the potential of our approach as a promising solution to overcome hardware impairments in RIS-aided wireless communication systems.

In this paper, we study the impact of pilot contamination in a system where two operators serve their respective users with the assistance of two wide-band reconfigurable intelligent surfaces (RIS), each belonging to a single operator. We consider one active user per operator and they use disjoint narrow frequency bands. Although each RIS is dedicated to a single operator, both users' transmissions are reflected by both RISs. We show that this creates a new kind of pilot contamination effect when pilots are transmitted simultaneously. Since combating inter-operator pilot contamination in RIS-assisted networks would require long pilot signal sequences to maintain orthogonality among the users of different operators, we propose the orthogonal configurations of the RISs. Numerical results show that this approach completely eliminates pilot contamination, and significantly improves the performance in terms of channel estimation and equalization by removing the channel estimation bias.

In the uplink of multiuser multiple input multiple output (MU-MIMO) systems operating over aging channels, pilot spacing is crucial for acquiring channel state information and achieving high signal-to-interference-plus-noise ratio (SINR). Somewhat surprisingly, very few works examine the impact of pilot spacing on the correlation structure of subsequent channel estimates and the resulting quality of channel state information considering channel aging. In this paper, we consider a fast-fading environment characterized by its exponentially decaying autocorrelation function, and model pilot spacing as a sampling problem to capture the inherent trade-off between the quality of channel state information and the number of symbols available for information carrying data symbols. We first establish a quasi-closed form for the achievable deterministic equivalent SINR when the channel estimation algorithm utilizes multiple pilot signals. Next, we establish upper bounds on the achievable SINR and spectral efficiency, as a function of pilot spacing, which helps to find the optimum pilot spacing within a limited search space. Our key insight is that to maximize the achievable SINR and the spectral efficiency of MU-MIMO systems, proper pilot spacing must be applied to control the impact of the aging channel and to tune the trade-off between pilot and data symbols.

The optimal power adaptation problem is investigated for vector parameter estimation according to various Fisher information based optimality criteria. By considering an observation model that involves a linear transformation of the parameter vector and an additive noise component with an arbitrary probability distribution, six different optimal power allocation problems are formulated based on Fisher information based objective functions. Via optimization theoretic approaches, various closed-form solutions are derived for the proposed problems. Also, the results are extended to cases in which nuisance parameters exist in the system model or certain types of nonlinear transformations are applied on the parameter vector. Numerical examples are presented to investigate performance of the proposed power allocation strategies.

We investigate the joint transmit beamforming and reconfigurable intelligent surface (RIS) configuration problem to maximize the sum downlink rate of a RIS-aided cellular multiuser multiple input single output (MU-MISO) system under imperfect channel state information (CSI) and hardware impairments by considering a practical phase-dependent RIS amplitude model. To this end, we present a novel deep reinforcement learning (DRL) framework and compare its performance against a vanilla DRL agent under two scenarios: the golden standard where the base station (BS) knows the channel and the phasedependentRIS amplitude model perfectly, and the mismatch scenario where the BS has imperfect CSI and assumes idealRIS reflections. Our numerical results show that the introduced framework substantially outperforms the vanilla DRL agent under mismatch and approaches the golden standard.