The increasing demand for wireless data traffic poses a significant challenge for current cellular networks, requiring each new technology generation to enhance network capacity and coverage, and spectral efficiency (SE) per connected device. Massive multiple-input multiple-output (MIMO) technology has emerged as a key component of 5G and leverages a large number of antennas at each access point (AP) to spatially multiplex many user equipments (UEs) over the same time-frequency resources. Looking beyond 5G, the new cell-free massive MIMO technology has gained considerable attention due to its ability to exploit spatial macro diversity and achieve higher interference resilience. Unlike traditional cellular networks, the cell-free architecture consists of a dense deployment of distributed APs that collaboratively serve UEs across a large coverage area without predefined cell boundaries. This architecture improves the mobile network coverage and aims to provide a more uniform quality of service throughout the network. However, the primary challenges of cell-free massive MIMO include the high computational complexity required for signal processing and the substantial fronthaul capacity needed for information exchange between APs. Moreover, another major challenge is handover management to cope with changing channel conditions and UE mobility; since in a cell-free network handover needs to consider how to dynamically evolve the serving set of APs to each UE, which is more complicated than in a cellular network where each UE is served by a single AP and handover means changing the serving AP.

In this doctoral thesis, we provide distributed solutions to research problems related to power allocation and mobility management to address some of the inherent challenges of the cell-free network architecture. Additionally, we introduce a new method for characterizing unknown interference in wireless networks. Moreover, we propose efficient optimization procedures in the context of multicast beamforming optimization and establish a novel method for rank reduction in conjunction with semidefinite relaxation (SDR).

For the problem related to power allocation, a distributed machine learning-based solution that provides a good trade-off between SE performance and applicability for implementation in large-scale networks is developed with reduced fronthaul requirements and computational complexity as compared to a centralized solution, where the power allocation for all APs is computed at a central processor. The solution is divided in a way that enables each AP, or group of APs, to separately decide on the power coefficients to the UEs based on the locally available information at the AP without exchanging information with the other APs, however, still attempting to achieve a network wide optimization objective. 

Regarding mobility management, a new soft handover procedure is devised for updating the serving sets of APs and assigning pilot signals to each UE in a dynamic scenario considering UE mobility. The algorithm is tailored to reduce the required number of handovers per UE and changes in pilot assignment. Numerical results show that our proposed solution identifies the essential refinements since it can deliver comparable SE to the case when the AP-UE association is completely redone.

As for interference modeling, we developed a new Bayesian-based technique to model the distribution of the unknown interference arising from scheduling variations in neighbouring cells. The method is shown to provide accurate statistical modeling of the unknown interference power and an effective tool for robust rate allocation in the uplink with a guaranteed target outage performance. The method was later extended to account for the unknown interference of neighbouring clusters in a cell-free network architecture.

Many wireless communication applications require sending the same data to multiple UEs; for example, in streaming live events, distributing software updates, or training of federated learning models. Physical-layer multicasting presents an efficient transmission topology to exploit the beamforming capabilities at the transmitting nodes and broadcast nature of the wireless channel to satisfy the demand for the same content from several UEs. The uniform service quality and improved coverage of the cell-free network architecture are particularly suitable for this transmission topology. In this regard, we propose a novel successive elimination algorithm coupled with SDR to extract a near-global optimal rank-1 beamforming solution to the max-min fairness (MMF) multicast problem in a cell-free massive MIMO network. A specifically tailored optimization algorithm is then designed, leveraging the alternating direction method of multipliers (ADMM) and offering significant improvements in computational requirements.

Den ökande efterfrågan på trådlös datatrafik utgör en betydande utmaning för dagens cellulära nätverk, vilket kräver att varje ny teknikgeneration förbättrar nätverkskapacitet och täckningen, samt spektraleffektivitet (SE) per uppkopplad enhet. Massiv MIMO-teknik (multiple-input multiple-output) har dykt upp som en viktig komponent i 5G då den använder ett stort antal antenner på varje accesspunkt (AP) för att rumsligt multiplexa många användarutrustningar (UE) över samma tidsfrekvensresurser. Om man ser bortom 5G så har den nya cellfria massiva MIMO-tekniken fått stor uppmärksamhet på grund av sin förmåga att utnyttja rumslig makrodiversitet och uppnå högre interferensmotståndskraft. Till skillnad från traditionella cellulära nätverk består den cellfria arkitekturen av en tät uppsättning av distribuerade AP:er som samarbetar för att betjäna UE:er över ett stort täckningsområde utan fördefinierade cellgränser. Denna arkitektur förbättrar mobilnätets täckning och syftar till att ge en mer enhetlig tjänstekvalitet i hela nätverket. De primära utmaningarna med cellfri massiv MIMO inkluderar den höga beräkningskomplexiteten som krävs för signalbehandling och den betydande fronthaul-kapacitet som behövs för informationsutbyte mellan AP:er. En annan stor utmaning är dessutom överräcknings-hantering för att klara av ändrade kanalförhållanden och UE-mobilitet; eftersom överräckning i ett cellfritt nätverk måste överväga hur man dynamiskt förändrar den betjänande uppsättningen av AP:er till varje UE, vilket är mer komplicerat än i ett cellulärt nätverk där varje UE betjänas av en enda AP och överräckning endast innebär att byta ansvarig AP.

I den här doktorsavhandlingen presenterar vi distribuerade lösningar på forskningsproblem relaterade till effektreglering och mobilitetshantering för att hantera några av de inneboende utmaningarna med den cellfria nätverksarkitekturen. Dessutom introducerar vi en ny metod för att karakterisera okända störningar i trådlösa nätverk. Vi föreslår även effektiva optimeringsprocedurer för optimering av multicast-lobformning och etablerar en ny metod för rangreduktion i samband med semidefinit relaxering (SDR).

För problemen kopplade till effektreglering utvecklas en distribuerad maskininlärningsbaserad lösning som ger en bra avvägning mellan SE-prestanda och tillämpbarhet för implementering i storskaliga nätverk med minskade fronthaulkrav och beräkningskomplexitet jämfört med en centraliserad lösning, där effektregleringen för alla AP:er beräknas på en central processor. Lösningen är uppdelat på ett sätt som gör det möjligt för varje AP, eller grupp av AP:er, att separat besluta om effektkoefficienterna till UE:er baserat på den lokalt tillgängliga informationen vid AP:erna utan att utbyta information med de övriga AP:erna, men ändå försöka uppnå ett nätverksomfattande optimeringsmål. 

När det gäller mobilitetshantering utformas en ny överräcknings-procedur för uppdatering av de betjänande uppsättningarna av AP:er och tilldelas pilotsignaler till varje UE i ett dynamiskt scenario med hänsyn till UE-mobilitet. Algoritmen är skräddarsydd för att minska det nödvändiga antalet överräckningar per UE och förändringar i pilottilldelningen. Våra numeriska resultat visar att den föreslagna lösningen identifierar de väsentliga förfiningarna eftersom den kan leverera jämförbar SE som i fallet där AP-UE-associationen görs om helt.

När det gäller störningsmodellering utvecklade vi en ny Bayesiansk metod för att modellera fördelningen av de okända störningar som uppstår på grund av schemaläggningsvariationer i närliggande celler. Metoden har visat sig ge korrekt statistisk modellering av störningseffekten och är ett effektivt verktyg för robust hastighetsallokering i upplänken med en garanterad avbrottsprestanda. Metoden utökades senare för att ta hänsyn till de okända störningarna från angränsande kluster i en cellfri nätverksarkitektur.

Många trådlösa kommunikationstillämpningar kräver att samma data skickas till flera UE:er, till exempel vid streaming av liveevenemang, distribution av programuppdateringar eller träning av federerade inlärningsmodeller. Multicasting på det fysiska lagret är en effektiv överföringsmetod som utnyttjar lobformningsförmågan vid AP:erna och spridningsförmågan hos den trådlösa kanalen för att tillfredsställa efterfrågan på samma innehåll från flera UE:er. Den enhetliga servicekvaliteten och förbättrade täckningen hos den cellfria nätverksarkitekturen är särskilt lämplig för denna överföringstopologi. I detta fall föreslår vi en ny gradvis elimineringsalgoritm kopplad till SDR för att extrahera en nästan globalt optimal rang-1-lobformningslösning för multicastproblemet med max-min-rättvisa i ett cellfritt massivt MIMO-nätverk. En skräddarsydd optimeringsalgoritm designas sedan för att utnyttja en metod som kallas alternating direction method of multipliers (ADMM) och erbjuder betydande förbättringar av beräkningsbehov.

Federated learning (FL) enables distributed model training by exchanging models rather than raw data, preserving privacy and reducing communication overhead. However, as the number of FL users grows, traditional wireless networks with orthogonal access face increasing latency due to limited scalability. Cell-free massive multiple-input multiple-output (CFmMIMO) networks offer a promising solution by allowing many users to share the same time-frequency resources. While CFmMIMO enhances energy efficiency through spatial multiplexing and collaborative beamforming, it remains crucial to adapt its physical layer operation to meticulously allocate uplink transmission powers to the FL users. To this aim, we study the problem of uplink power allocation to maximize the number of global FL iterations while jointly optimizing uplink energy and latency. The key challenge lies in balancing the opposing effects of transmission power: increasing power reduces latency but increases energy consumption, and vice versa. Therefore, we propose two power allocation schemes: one minimizes a weighted sum of uplink energy and latency to manage the trade-off, while the other maximizes the achievable number of FL iterations within given energy and latency constraints. We solve these problems using a combination of Brent's method, coordinate gradient descent, the bisection method, and Sequential Quadratic Programming (SQP) with BFGS updates. Numerical results demonstrate that our proposed approaches outperform state-of-the-art power allocation schemes, increasing the number of achievable FL iterations by up to 62%, 93%, and 142% compared to Dinkelbach, max-sum rate, and joint communication and computation optimization methods, respectively.

Federated learning (FL) is a distributed learning paradigm wherein users exchange FL models with a server instead of raw datasets, thereby preserving data privacy and reducing communication overhead. However, the increased number of FL users may hinder completing large-scale FL over wireless networks due to high imposed latency. Cell-free massive multiple-input multiple-output (CFmMIMO) is a promising architecture for implementing FL because it serves many users on the same time/frequency resources. While CFmMIMO enhances energy efficiency through spatial multiplexing and collaborative beamforming, it remains crucial to meticulously allocate uplink transmission powers to the FL users. In this paper, we propose an uplink power allocation scheme in FL over CFmMIMO by considering the effect of each user's power on the energy and latency of other users to jointly minimize the users' uplink energy and the latency of FL training. The proposed solution algorithm is based on the coordinate gradient descent method. Numerical results show that our proposed method outperforms the well-known max-sum rate by increasing up to 27% and max-min energy efficiency of the Dinkelbach method by increasing up to 21 % in terms of test accuracy while having limited uplink energy and latency budget for FL over CFmMIMO.

Physical layer multicasting is an efficient transmission technique that exploits the beamforming potential at the transmitting nodes and the broadcast nature of the wireless channel, together with the demand for the same content from several UEs. This paper addresses the max-min fair multigroup multicast beamforming optimization, which is an NP-hard problem. We propose a novel iterative elimination procedure coupled with semidefinite relaxation (SDR) to find the near-global optimum rank-1 beamforming vectors in a cell-free massive MIMO (multiple-input multiple-output) network setup. The proposed optimization procedure shows significant improvements in computational complexity and spectral efficiency performance compared to the SDR followed by the commonly used randomization procedure and the state-of-the-art difference-of-convex approximation algorithm. The significance of the proposed procedure is that it can be utilized as a rank reduction method for any problem in conjunction with SDR.

This paper considers a mmWave cell-free massive MIMO (multiple-input multiple-output) network composed of a large number of geographically distributed access points (APs) simultaneously serving multiple user equipments (UEs) via coherent joint transmission. We address UE mobility in the downlink (DL) with imperfect channel state information (CSI) and pilot training. Aiming at extending traditional handover concepts to the challenging AP-UE association strategies of cell-free networks, distributed algorithms for joint pilot assignment and cluster formation are proposed in a dynamic environment considering UE mobility. The algorithms provide a systematic procedure for initial access and update of the serving APs and assigned pilot sequence to each UE. The principal goal is to limit the necessary number of AP and pilot changes, while limiting computational complexity. We evaluate the performance, in terms of spectral efficiency (SE), with maximum ratio and regularized zero-forcing precoding. Results show that our proposed distributed algorithms effectively identify the essential AP-UE association refinements with orders-of-magnitude lower computational time compared to the state-of-the-art. It also provides a significantly lower average number of pilot changes compared to an ultra-dense network (UDN). Moreover, we develop an improved pilot assignment procedure that facilitates massive access to the network in highly loaded scenarios.

Co-channel interference poses a challenge in any wireless communication network where the time-frequency resources are reused over different geographical areas. The interference is particularly diverse in cell-free massive multiple-input multiple-output (MIMO) networks, where a large number of user equipments (UEs) are multiplexed by a multitude of access points (APs) on the same time-frequency resources. For realistic and scalable network operation, only the interference from UEs belonging to the same serving cluster of APs can be estimated in real-time and suppressed by precoding/combining. As a result, the unknown interference arising from scheduling variations in neighboring clusters makes the rate adaptation hard and can lead to outages. This paper aims to model the unknown interference power in the uplink of a cell-free massive MIMO network. The results show that the proposed method effectively describes the distribution of the unknown interference power and provides a tool for rate adaptation with guaranteed target outage.

The existence of unknown interference is a prevalent problem in wireless communication networks. Especially in multi-user multiple-input multiple-output (MIMO) networks, where a large number of user equipments are served on the same time-frequency resources, the outage performance may be dominated by the unknown interference arising from scheduling variations in neighboring cells. In this letter, we propose a Bayesian method for modeling the unknown interference power in the uplink of a cellular network. Numerical results show that our method accurately models the distribution of the unknown interference power and can be effectively used for rate adaptation with guaranteed target outage performance.

This paper considers a cell-free massive multiple-input multiple-output (MIMO) system that consists of a large number of geographically distributed access points (APs) serving multiple users via coherent joint transmission. The downlink performance of the system is evaluated, with maximum ratio and regularized zero-forcing precoding, under two optimization objectives for power allocation: sum spectral efficiency (SE) maximization and proportional fairness. We present iterative centralized algorithms for solving these problems. Aiming at a less computationally complex and also distributed scalable solution, we train a deep neural network (DNN) to approximate the same network-wide power allocation. Instead of training our DNN to mimic the actual optimization procedure, we use a heuristic power allocation, based on large-scale fading (LSF) parameters, as the pre-processed input to the DNN. We train the DNN to refine the heuristic scheme, thereby providing higher SE, using only local information at each AP. Another distributed DNN that exploits side information assumed to be available at the central processing unit is designed for improved performance. Further, we develop a clustered DNN model where the LSF parameters of a small number of APs, forming a cluster within a relatively large network, are used to jointly approximate the power coefficients of the cluster.

This paper addresses mobility management in the downlink of a mmWave cell-free massive MIMO (multiple-input multiple-output) network with imperfect channel knowledge obtained from pilot training. The network consists of a large number of geographically distributed access points (APs) simultaneously serving multiple user equipments (UEs) via coherent joint transmission. The objective is to extend traditional handover concepts to the challenging AP-UE association strategies of cell-free networks. To this end, we propose a distributed algorithm for joint pilot assignment and cluster formation in a dynamic environment considering UE mobility. The primary goal is to limit the necessary number of AP and pilot changes, with reasonable computational complexity. We evaluate the performance in terms of the spectral efficiency with maximum ratio and regularized zero-forcing precoding. Results show that our proposed distributed algorithm effectively identifies the essential AP-UE association refinements. Moreover, it provides a significantly lower average number of pilot changes compared to an ultra-dense network.

The ever-growing demand of wireless traffic poses a challenge for current cellular networks. Each new generation must find new ways to boost the network capacity and spectral efficiency (SE) per device. A pillar of 5G is massive multiple-input-multiple-output (MIMO) technology. Through utilizing a large number of antennas at each transmitting node, massive MIMO has the ability to multiplex several user equipments (UEs) on the same time-frequency resources via spatial multiplexing. Looking beyond 5G, cell-free massive MIMO has attracted a lot of attention for its ability to utilize spatial macro diversity and higher resilience to interference. The cell-free architecture is based on a large number of distributed access points (APs) jointly serving the UEs within a coverage area without creating artificial cell boundaries. It provides a promising solution that is focused on delivering uniform service quality throughout the mobile network. The main challenges of the cell-free network architecture lie in the computational complexity for signal processing and the huge fronthaul requirements for information exchange among the APs.

In this thesis, we tackle some of the inherent problems of the cell-free network architecture by providing distributed solutions to the power allocation and mobility management problems. We then introduce a new method for characterizing unknown interference in wireless networks.

For the problem of power allocation, a distributed learning-based solution that provides a good trade-off between SE performance and applicability for implementation in large-scale networks is developed with reduced fronthaul requirements and computational complexity. The problem is divided in a way that enables each AP (or group of APs) to separately decide on the power coefficients to the UEs based on the locally available information at the AP without exchanging information with the other APs, however, still attempting to achieve a network wide optimization objective. 

Regarding mobility management, a handover procedure is devised for updating the serving sets of APs and assigned pilot to each UE in a dynamic scenario considering UE mobility. The algorithm is tailored to reduce the required number of handovers per UE and changes in pilot assignment. Numerical results show that our proposed solution identifies the essential refinements since it can deliver comparable SE to the case when the AP-UE association is completely redone.

Finally, we developed a new technique based on a Bayesian approach to model the distribution of the unknown interference arising from scheduling variations in neighbouring cells. The method is shown to provide accurate modelling for the unknown interference power and an effective tool for robust rate allocation in the uplink with a guaranteed target outage performance.

Den ständigt växande efterfrågan på trådlös datatrafik är en stor utmaning för dagens mobilnät. Varje ny nätgeneration måste hitta nya sätt att öka den totala kapaciteten och spektraleffektiviteten (SE) per uppkopplad enhet. En pelare i 5G är massiv-MIMO-teknik (multiple-input-multiple-output). Genom att använda ett stort antal antenner på varje mobilmast har massiv MIMO förmågan att kommunicera med flera användarutrustningar (eng. user equipment, UE) på samma tid/frekvensresurser via så kallad rumslig multiplexing. Om man ser bortom 5G-tekniken så har cellfri massiv-MIMO väckt stort intresse tack vare sin förmåga att utnyttja rumslig makrodiversitet för att förbättra täckningen och uppnå högre motståndskraft mot störningar. Den cellfria arkitekturen bygger på att ha ett stort antal distribuerade accesspunkter (AP) som gemensamt serverar UE:erna inom ett täckningsområde utan att dela upp området konstgjorda celler. Detta är en lovande lösning som är fokuserad på att leverera enhetliga datahastigheter i hela mobilnätet. De största forskningsutmaningarna med den cellfria nätverksarkitekturen ligger i beräkningskomplexiteten för signalbehandling och de enorma kraven på fronthaul-kablarna som möjliggör informationsutbyte mellan AP:erna.

I den här avhandlingen löser vi några av de grundläggande utmaningarna med den cellfria nätverksarkitekturen genom att tillhandahålla distribuerade algoritmlösningar på problem relaterade till signaleffektreglering och mobilitetshantering. Vi introducerar sedan en ny metod för att karakterisera okända störningar i trådlösa nätverk.

När det gäller signaleffektreglering så utvecklas en distribuerad inlärnings-baserad metod som ger en bra avvägning mellan SE-prestanda och tillämpbarhet för implementering i storskaliga cellfria nätverk med reducerade fronthaulkrav och lägre beräkningskomplexitet. Lösningen är uppdelat på ett sätt som gör det möjligt för varje AP (eller grupp av AP) att separat besluta om effektkoefficienterna relaterade till varje UE baserat på den lokalt tillgängliga informationen vid AP:n utan att utbyta information med de andra AP:erna, men ändå försöka uppnå ett nätverksomfattande optimeringsmål.

När det gäller mobilitetshantering utformas en överlämningsprocedur som dynamiskt uppdaterar vilken uppsättning av AP:er som servar en viss UE och vilken pilotsekvens som används när den rör sig över täckningsområdet. Algoritmen är skräddarsydd för att minska antalet överlämningar per UE och förändringar i pilottilldelningen. Numeriska resultat visar att vår föreslagna lösning identifierar de väsentliga förfiningarna eftersom den kan leverera jämförbar SE som när AP-UE-associationen görs om helt och hållet.

Slutligen utvecklade vi en ny Bayesiansk metod för att modellera den statistiska fördelningen av de okända störningarna som uppstår på grund av schemaläggningsvariationer i närliggande celler. Metoden har visat sig ge en korrekt modell av den okända störningseffekten och är ett effektivt verktyg för robust SE-allokering i upplänken med en garanterad maximal avbrottsnivå.