In Device-to-Device (D2D) enabled cellular networks, user-to-network relaying can be handled for improving the performance of cellular networks. When relays are in mobility, a dynamic relay selection strategy is unavoidable. In this paper, we propose a dynamic policy of relay selection that maximizes the performance of cellular networks (e.g. throughput, reliability, coverage) under cost constraints (e.g. transmission power, power budget). We model the relays' dynamics as a Markov Decision Process (MDP). Since only the locations of the selected relays are observed, the sequential relay selection process is formulated as a Constrained Partially Observable Markov Decision Process (CPOMDP). The exact solution of such framework is intractable to find, therefore we prove the submodularity property of the reward and cost functions and deduce a greedy point based value iteration solution. Considering the throughput as reward metric and the energy as cost metric, numerical results are illustrated to endorse the proposed relay selection policy and to show how introducing D2D relaying can highly improve the performance of cellular networks.

Device-to-device (D2D) communications is a promising technique for improving the efficiency of 5G networks. Employing channel adaptive resource allocation can yield to a large enhancement in almost any performance metric of D2D communications (e.g. Energy Efficiency). Centralized approaches require the knowledge of D2D links' Channel State Information (CSI) at the BS level. However, CSI reporting suffers from the limited number of resources available for feedback transmission. Alternately, we propose a distributed algorithm for resource allocation that benefits from the users' knowledge of their local CSI in order to minimize the users' transmission power while maintaining predefined throughput constraint. The key idea is that users compute their local performance metrics (e.g. energy efficiency) and then use a new signaling mechanism to share these values between each other. Under some condition, the performance of this distributed algorithm achieves that of theideal scheduling (i.e. with a global CSI knowledge of all the D2D links). We describe how this technique can be simply implemented by adapting existing CSI reporting (e.g. in Long-Term Evolution (LTE) systems). Furthermore, numerical results are presented to corroborate our claims and demonstrate the gain that the proposed distributed scheduling brings to cellular networks compared to the bestcentralized-limited feedback scheduling.

The main weakness of coordination techniques in LTE-Advanced networks is the extra resource consumption incurred by the joint transmission from several base stations. In this paper, we propose a new scheduling policy that performs coordination primarily for users staying at the cell edge, without mobility. Other cell-edge users are likely to move and to be served in better radio conditions where cell coordination is not required. We compare the performance of this algorithm to other usual scheduling policies in the presence of elastic traffic through the analysis of flow-level traffic models.

It is well-known that opportunistic schedulers can exploit multi-user diversity and channel variations in the presence of elastic traffic to improve users' QoE. As more and more adaptive streaming traffic appear in the wireless system, it is needed to verify the performance impacts of deploying opportunistic schedulers and other policies. This paper examines the video bit rate and the buffer surplus in various scenarios of mobility under several scheduling policies, and proposes for operators to deploy round-robin policy in the static scenario. In the cases with users' mobility, there are performance trade-offs between the video bit rate and the buffer surplus and scheduling policy is suggested based on the targeted metric. In order to obtain an intermediate results, discriminatory scheduling is also investigated and suggested in this case.

This paper investigates the network offloading capability of Device to Device (D2D) communication when combined with content caching. We consider a mobile network where a proportion of users are equipped with caches from which other users can retrieve contents using direct D2D communications. In this case, the offloading capacity depends both on the cache performance and the quality of the D2D links. As of the cache performance, we develop an analytical model for the cache hit rate for different content prefetching strategies. We then inject this hit rate in a queuing theory analysis of the network capacity where the macro base station and the D2D links are modeled as coupled Processor Sharing (PS) queues. We develop closed-form solutions for the stability region and the performance gains for both one-hop and two-hop D2D. Our simulation results illustrate the offloading capacity and present optimal cache and network parameters that maximize the benefit of D2D communications.

Optimization of mobile networks is a crucial task for operators aiming to offer a high quality of service. In this paper, we focus on the handover (HO) problem. An HO is the procedure through which a user changes its serving cell, ensuring service continuity for mobile users. The HO procedure is typically governed by several variables: the received signals powers and the HO parameters (offsets and hysteresis) corresponding to the different cells. HO parameters allow to modify the cell borders in order to balance loads, and prevent radio link failures and HO ping-pongs. There are two approaches for defining HO parameters: one is to use per-cell parameters and the other is to use per-cell-pair parameters. Per-cell-pair parameters provide more degrees of freedom in HO parameter optimization as we benefit from independent parameters for each neighbour. However, we show that it can lead to instabilities in the form of continuous chained HO, i.e., several HOs that follow one shortly after another, among three or more cells. Then, we show that using the same HO parameters toward all neighbouring cells, i.e., per-cell parameters, is risk free of such problems. We provide numerical results outlining the tendencies of this risk.

Coverage optimization is an important process for the operator, as it is a crucial prerequisite toward offering a satisfactory quality of service to the end users. The first step of this process is coverage prediction, which can be performed by interpolating geo-located measurements reported to the network by mobile user's equipments. In the previous works, we proposed a low complexity coverage prediction algorithm based on the adaptation of the geo-statistics fixed rank kriging (FRK) algorithm. We supposed that the geo-location information reported with the radio measurements was perfect, which is not the case in reality. In this paper, we study the impact of location uncertainty on the coverage prediction accuracy and we extend the previously proposed algorithm to include geo-location error in the prediction model. We validate the proposed algorithm using both simulated and real-field measurements. The FRK is extended to take into account that the location uncertainty proves to enhance the prediction accuracy while keeping a reasonable computational complexity.

An important problem of today's mobile network operators is to bring down the capital expenditures and operational expenditures. One strategy is to automate the parameter tuning on the small cells through the so-called self-organizing network (SON) functionalities, such as cell range expansion, mobility robustness optimization, or enhanced Inter-Cell Interference Coordination. Having several of these functionalities in the network will surely create conflicts, as, for example, they may try to change the same parameter in the opposite directions. This raises that the need for an SON COordinator (SONCO) meant to arbitrate the parameter change requests of the SON functions, ensuring some degree of fairness. It is difficult to anticipate the impact of accepting several simultaneous requests. In this paper, we provide a SONCO design based on reinforcement learning (RL) as it allows us to learn from previous experiences and improve our future decisions. Typically, RL algorithms are complex. To reduce this complexity, we employ two flavors of function approximation and provide a study-case. Results show that the proposed SONCO design is capable of biasing this fairness among the SON functions by means of weights attributed to the SON functions. Also, we evaluate the tracking capability of the algorithms.

In this paper, we assess the performance of inter-cell coordination in the presence of mobility. This performance depends primarily on the resource allocation scheme. Indeed, a scheduling strategy which may seem efficient when users are static can lead to bad performance when users are mobile. Several scheduling policies are investigated. Their performance critically depends on their ability to predict users' mobility. The results are based on the analysis of flow-level traffic models.

In trying to meet the demands of traffic hungry users, mobile network operators are faced with increased CAPital EXpenditures (CAPEX) and OPerational EXpenditures (OPEX). The Self Organizing Network (SON) functions have been introduced by 3GPP as a means to cut down these costs. There are mainly 3 categories of such functions: self-configuration, self-optimization and self-healing. In this paper we focus on the second which represents the SON functions performing a runtime optimization of the network. We center our attention on LTE heterogeneous networks. Having several SON functions in a network may lead to conflicts and potentially to bad network Key Performance Indicators (KPIs). Thus a troubleshooting mechanism has to be envisaged. Such a mechanism typically contains 3 steps: fault detection, cause diagnosis and solution deployment. In this paper we tackle the first two and we study the feasibility of using the Naive Bayes Classifier (NBC) in order to build a framework for SON Conflict Diagnosis (SONCD). We provide numeric results proving the feasibility of the framework.

In this paper, we assess the performance gains of mobility on the downlink of cellular data networks. These gains are only due to the elastic nature of traffic and thus observed even under a blind, fair scheduling scheme: data are more likely transmitted when users are close to the base stations, in good radio conditions. This phenomenon is further amplified by opportunistic scheduling schemes that exploit multi-user diversity. The results are based on the analysis of flow-level traffic models and validated by system-level simulations.

The continuously increasing traffic demand faces us with increased CAPital Expenditures (CAPEX) and Operational Expenditure (OPEX). Self Organizing Network (SON) functions aim to lower these costs by automating the network tuning. A SON instance is a realization of a SON function which can tune one or a set of cells. Having several uncoordinated SON functions in the network creates a risk for conflicts and instability. This raises the need for a SON Coordinator (SONCO) meant to deal with these issues. In this work we consider that on each cell we have one SON instance of each SON function. We present the design of a SONCO which arbitrates conflicts based on weights attributed to the SON functions. The design makes use of Reinforcement Learning (RL) with function approximation. We provide a low complexity approximation of the action-value function based on a number of parameters that scales linearly with the number of cells. We present a study case with the Mobility Load Balancing (tuning the Cell Individual Offset (CIO)) and Mobility Robustness Optimization (tuning the CIO and the handover hysteresis) functions, where the SONCO deals with the conflicts on the CIOs. Numerical results prove that we can orchestrate the SON functions through SONCO configurations that reflect different operator policies.

Coverage optimization is a crucial task for a radio network operator. An accurate coverage estimation is a key prerequisite for efficient coverage analysis and optimization. In this paper, we propose a coverage prediction method based on statistical modeling of the wireless environment. We build a Radio Environment Map by interpolating geo-located measurements using the Kriging spatial prediction technique. Moreover, as we perform Kriging on massive observation datasets obtained through field measurement campaigns, we use Fixed Rank Kriging, to reduce the complexity of the Kriging algorithm. We apply the FRK algorithm for Long Term Evolution (LTE) network coverage prediction. We consider as observation data, the coverage measurements obtained by operational drive tests in a rural area. Numerical results show that by using the FRK algorithm, we fulfill a good trade-off between computational complexity and prediction accuracy.

With the emergence of Self-Organizing Network (SON) functions network operators are faced with a practical problem: coordination of SON instances. The SON functions are usually designed in a standalone manner, i.e. they do not take into account the possibility that other instances of the same or different SON functions may be running in the network. This creates the risk of conflicts and network instability. Therefore a SON COordinator (SONCO) is needed. In this paper we design an operator centric SONCO that sees the SON instances as black-boxes, i.e. it does not know the algorithm inside the SON functions. Our aim is to improve the network stability (i.e. number of parameter changes) for SON instances of the same SON function. We employ Reinforcement Learning (RL) in order to profit from the information on the past SONCO decisions. We simplify the expression of the action-value function and we use state aggregation to further reduce the required state space, making it scale linearly with the number of coordinated cells. We provide a study case with the Mobility Load Balancing (MLB) function independently instantiated on every cell. The results show that the proposed SONCO improves the network stability.

In Long Term Evolution(LTE) networks one of the main focuses is on automating the network optimization. This is done through so called Self Organizing Network (SON) functions like Mobility Load Balancing (MLB), Mobility Robustness Optimization(MRO) and others. A SON instance is a realization of a SON function that governs (optimizes) one or a cluster of eNBs. The SON functions are built in a standalone manner, i.e. without considering the existence of other SON instances. So they do not necessarily operate in a coordinated fashion, especially in a network where different SON instances may come from different vendors. Thus we face a risk of generating conflicts and instabilities in the network and so this raises the need for a SON COordinator (SONCO) . The SONCO, built from an operator point of view, sees the SON instances as black boxes and has a very limited amount of information on them. In these conditions the SONCO has to solve conflicts and improve the network stability. In this paper we propose a Reinforcement Learning (RL) based solution for coordinating SON instances that run independently on neighboring eNBs and we provide results for a case study with MLB instances. We analyze the scalability of our solution and we provide numerical results showing how improvements in network stability can be obtained.

During the last decade a lot of effort has been spent on cellular network optimization to improve network capacity and end-user Quality of Service (QoS). Coverage analysis remains as one of the essential topics on which mobile operators still need innovation in terms of performance and cost. Manual coverage analysis is an inefficient and costly task. Radio Environment Maps (REMs) is an efficient coverage analysis solution for present-day cellular networks. REM concept consists of spatially interpolating geo-located measurements to build the whole coverage map using a spatial interpolation technique originating from geo-statistics. Kriging is such a powerful technique which results in high performance in terms of prediction quality. However, this method is costly in terms of computational complexity especially for large datasets: computational complexity of Kriging is O(n3) where n is the number of measurements. This paper proposes the application of a variant of Kriging, Fixed Rank Kriging (FRK), to coverage analysis in order to reduce the computational complexity of the spatial interpolation while keeping an acceptable prediction error.

Optimization of radio access networks is a challenge for cellular operators. With the deployment of new radio access technologies and advanced features, network optimization needs to be automated for performance improvement and OPerational EXpenditures (OPEX) reduction. In this paper, we propose a method for centralized self-optimization of pilot powers to perform load balancing between base stations. The proposed method uses a stochastic surrogate function to model the functional relationships between noisy Key Performance Indicators (KPIs) and network parameters, and subsequently performs optimization using a pattern search algorithm in a recursive manner. The method is applied to solve a high dimensional optimization problem of pilot power based load balancing in an LTE network. Results obtained using a flow level analytical network simulator show the advantages of the proposed recursive modeling and optimization approach as a promising solution in terms of robustness to noisy data and fast convergence to the optimum point for a high dimensional network optimization.

A major challenge to the successful operation and maintainability of LTE-A network is to address competing objectives such as improved capacity, reduced operational costs and complexity. In this paper we use a surrogate function to model the functional relationships between noisy Key Performance Indicators (KPIs) and Radio Resource Management (RRM) parameters and subsequently use a pattern search algorithm for optimization under constraints. The proposed methodology is applied to solve a self-optimization problem for eICIC in an LTE-A HetNet deployment. Results obtained using a system level simulator show promising results for faster convergence towards optimal network configuration.

Deployment of Long Term Evolution (LTE) amp; LTE- Advanced networks will be challenged by cost and complexity. Self Organizing Network (SON) functionalities promise significant improvement of the network in terms of reducing OPerational EXpenditure (OPEX) and performance improvement. In this paper, we propose a recursive automated optimization method which builds statistical models of the functional relationships between noisy Key Performance Indicators (KPIs) and network parameters; and performs stochastic optimization during the model building process. The proposed methodology is applied to a centralized intra-LTE Mobility Load Balancing (MLB) problem and its performance is evaluated through system level simulations. The results show that the proposed modeling and optimization approach is a promising solution for centralized intra-LTE MLB in terms of optimization performance and convergence under noisy data.

The development of self-management solutions for (multi-technology, multi-layer) mobile communication networks is driven by their increasing operational complexity. Initial stand-alone SON (Self-Organizing Networks) solutions are already available, but are not sufficient to handle the networks of tomorrow. In this paper we present our approach at developing a unified management framework that integrates the existing and future advanced SON functions across several radio access technologies. The envisioned self-management system comprises (i) an integrated SON management system, in charge of policy transformation/supervision and conflict detection/handling; (ii) advanced multi-RAT/layer SON functions; and (iii) a Decision Support System providing measurement-based assistance for residual operational tasks, such as timely recommendations for targeted new site deployments.

In this contribution a layered radio environment map (REM) architecture is designed and applied in the framework of radio-access network optimization for heterogeneous LTE systems that comprise both macrocells and femtocells. We define layer as the hierarchical representation of a geographical area. In each layer, different instances of the same architectural block will have different spatial/temporal granularities, associated with network functionalities. The effectiveness of the proposed architecture to support LTE functions like automatic neighbor relation (ANR) and minimization of drive tests (MDT) is discussed. In addition, we present the benefits of using such architecture in the implementation of these functions along with its potential to bring performance gains.

The Self Organizing Network (SON) concept is a promising paradigm aiming to reduce both the capital and operational expenditures (CAPEX/OPEX) in future wireless networks through efficient self-configuration, self-optimization and self-healing functionalities. As for self-optimization, much effort has been devoted to standalone SON control loops, which, upon network metric observation, dynamically adjust the appropriate network parameters to enhance the overall network performance. However, less explored simultaneous operation of several SON control loops may cause conflicts between parameter adjustment and/or metric performance outcome. This paper focuses on SON mechanism interactions, providing a framework for further understanding and analysis. A basic two-SON mechanism interaction scenario will be provided along with an initial assessment of a simple SON interaction control.

In this paper we propose a Bayesian approach for estimating parameters of the radio propagation model, and an iterative Kriging interpolation algorithm for choosing the best candidate measurement to be retrieved into the Radio Environment Map (REM). We compare the performance with a random choice of the candidate measurement and show that our algorithm reduces the amount of measurement needed by 33%. The proposed algorithm has also the merit of being fast enough to be implemented in an online fashion for REMs with a grid size of 25m and for pedestrian mobile speeds.

Troubleshooting of wireless networks is a challenging network management task. We have developed, in a previous work, a new troubleshooting methodology, which we named Statistical Learning Automated Healing (SLAH). This methodology uses statistical learning, in particular logistic regression, to extract the functional relationships between the noisy Key Performance Indicators (KPIs) and Radio Resource Management (RRM) parameters. These relationships are then processed by an optimization engine so as to calculate the optimized RRM parameters which improve the KPIs of a degraded cell. The process is iterative and converges to the optimum RRM parameter value in few iterations, which makes it suitable for wireless networks. The present work focuses on the adaptation of SLAH for troubleshooting the mobility parameter, namely the handover margin, in 3G Long Term Evolution (LTE) networks. The simulation results, which we obtain for a practical use case, show the advantage of this new, automated troubleshooting methodology.

Abstract In this paper, a machine learning approach is proposed for automatic recognition of the radio access technology (RAT) based on supervised classification of spectrum power measurements in a cognitive radio context. Recognition of the active local RATs allows the cognitive terminal to make informed decisions on which wireless technology to camp, and to adapt its transmission to the wireless environment. To this end, supervised classification of signals obtained from wide band spectrum power measurements with low temporal resolution is carried out. As a first step, generic features are extracted from the power spectrum measurements. Then several supervised classification algorithms, namely the support vector machines (SVMs),k‐nearest neighbors (kNN) and C4.5 are tested. The effects of reducing the number of features on classification accuracy is also tested through feature selection methods. Finally, the impact of the signal duration on classification accuracy is investigated. The obtained results demonstrate that the tested supervised classification algorithms, combined with simple feature extraction and selection methods, are good candidates for RAT recognition from low temporal resolution spectrum power measurements. Copyright © 2009 John Wiley & Sons, Ltd.

In the context of cognitive radios, we present a method to sense wideband radio spectra from randomly distributed samples with an average sampling rate smaller than the Nyquist sampling rate. The method finds its roots in the compressed sensing paradigm. The Gabor time-frequency is employed as a sparsifying transform to represent the radio signals generated in the framework of TDMA/FDMA based radio access technologies. Indeed, many white spaces are observable in these wideband spectrograms. In addition, noisy measurements are modeled through the Basis Pursuit DeNoise formulation, which enables to reconstruct the time-frequency representation with some errors or information losses. The method is experimented on measured baseband radio signals sampled at 51.2MHZ and acquired in the GSM 900 downlink band. Finally, results about the effects of the algorithm parameters and about the algorithm complexity are provided and discussed.

Self organizing network (SON) functionalities are currently developed to improve network performance and management tasks. SON functionalities require efficient utilization of data extracted from the network. In this context, the paper has two objectives. First it is shown that one can use simple statistical learning techniques such as regression to extract a model from data. The model comprises closed form expressions that approximate the functional relations between measured key performance indicators (KPIs) and radio resource management (RRM) parameters. Second, it is shown how the model can be integrated in a monitoring process and be used to devise an efficient auto-tuning algorithm. To this end, two case studies of handover monitoring and handover auto-tuning in a LTE network are described and illustrate the application of the proposed approach.

In this work, we aim at finding the low dimensional hidden structures or manifolds that exist in the high dimensional data produced by a Radio Access Network (RAN). Specifically, we consider the Key Performance Indicators (KPIs) of a UMTS network. The KPI data is obtained by performing semi-dynamic simulations of a Radio Network Planning (RNP) tool. The low-dimensional manifold, yielding a meaningful and tractable representation of the performance indicators, facilitates the complicated tasks like monitoring, troubleshooting, fault detection, design, radio resource management etc. We have applied one second-order linear (PCA), one high-order linear (ICA) and one nonlinear technique (ISOMAP) of manifold learning and compared the results.

In this paper, we propose a method for predicting traffic values from spectrum measurements. The purpose is to obtain traffic information from the wireless environment for making informed decisions in order to balance traffic loads. We propose a supervised prediction scheme trained by real measurements, such that the method adapts to new radio contexts for new measurements. As a first step, features are extracted from spectrum measurements using probability distributions and a Principal Component Analysis. Then, the relevant features are selected based on a correlation criterion. In the third stage, the predictor, mapping the selected features to a traffic value, is trained using thekNearest Neighbours algorithm. The predictor reliability is then evaluated using a cross validation. The results obtained with spectrum measurements realized on the GSM900 and the DCS1800 frequency bands show that the proposed scheme achieves a reliable traffic prediction.

The cognitive radio paradigm is based on the ability of sensing the radio environment in order to make informed decisions. This paper describes the effects of sensing on the cognitive radio channels capacity region. Sensing is modeled as a compression channel, which results in partial knowledge of the primary messages at the cognitive transmitter. This model enables to impose constraints on the sensing strategy. First, the dirty paper channel capacity is derived when the channel encoder knows partially the side information. Then, the capacity area of the Gaussian cognitive channel with partial information is derived. Finally, numerical results illustrate the capacity reduction associated with constrained sensing, in comparison to the capacity of the cognitive radio channel.

This paper presents an online tuning approach for the ad-hoc reinforcement learning algorithms which are used for solving the exploitation-exploration dilemma of the opportunistic spectrum access, in dynamic environments. These algorithms originate from a well-known problem in computer science: the multi-armed bandit (MAB) problem and they have provided evidence to be viable solutions for the detection and exploration of white spaces in opportunistic spectrum access. Previous work (A. Ben Hadj Alaya-Feki et al., 2008) has shown that the reinforcement learning solutions of the MAB problem are very sensitive to the statistical properties of the wireless medium access and therefore need careful tuning according to the dynamic variations of the wireless environment. This paper deals with the online tuning of those algorithms by proposing and assessing two different approaches: 1-a meta learning approach where a second learner (meta learner) is used to learn the parameters of the base learner, and 2-the Exp3 algorithm that has been previously proposed for dynamical tuning of MAB parameters in other contexts. The simulation results obtained on an IEEE 802.11medium access scenario show that one of the proposed meta-learning methods, namely the change point detection method, achieves much better performance compared to the other methods.

In this paper, we propose a method for automatically extracting distinct spectral states present in spectrum power measurements. The purpose is to gain information on the wireless environment for constructing efficient policies and decisions in a flexible radio context. To achieve this, we perform unsupervised classification on data obtained from spectrum power measurements on the GSM 1800 downlink band. The number of spectral states having different spectral characteristics is automatically detected by a well-known efficient rate-distortion criterion. Spectral signals that constitute the power spectrum measurements are classified into these distinct spectral states through the k-means clustering algorithm. In order to eliminate redundancies and to decrease the computational complexity, Principal Component Analysis is performed on the set of spectral signals before classification. The time-averaged frequency profile of each spectral state can be considered as the quantization or compression of the spectral information contained by the spectral signal belonging to that spectral state. The preliminary results obtained on the GSM 1800 downlink spectrum measurements reveal that the proposed method successfully extracts the distinct spectral states of the wireless environment.

This paper introduces interference cartography, a simple and effective concept that helps detect, identify and use spectrum opportunities in a secondary spectrum usage context. Interference cartography combines measurements performed by different network entities (mobile terminals, base stations, access points) with the geo-location information and applies simple and effective spatial interpolation techniques to achieve a map which indicates the level of interference experienced at each mesh over the area of interest. Using this information, a secondary network can detect the presence of a primary network (or of other secondary networks) and can use spectrum opportunities without causing harmful interference to them. As an example, a reliable spatial interpolation technique, kriging, is applied to interference data obtained from a radio network simulator. Obtained results demonstrate that interference cartography is a promising concept that can enhance the performance of secondary spectrum usage.

Opportunistic spectrum access (OSA) is a hot topic in cognitive radio context. The main challenge of the OSA is to define improved spectral usage schemes, through the utilization of frequency holes in licensed bands. The multi armed bandit (MAB) is a reinforcement learning technique that can provide the secondary user with the adequate rules, in order to perform simultaneously 1) the exploitation of its external environment and 2) the exploration of the accumulated knowledge by transmitting in the identified white spaces. However, the MAB is sensitive to the non-stationarity of the channels' statistical characteristics. Thus, in this paper, we address an offline sensitivity study to optimize the parameter tuning in the used MAB allocation strategies. Also, we propose a semi dynamic parameter tuning scheme to achieve an online update of the MAB parameters. This adaptive MAB solution enhances the performance of the secondary user in dynamic environments.

This paper presents a constructive learning system that enhances the reliability and precision of radio network predictions. This task is achieved by finding a correspondence between the probability density distributions of simulated predictions and real measurement data collected from the radio network. Once this correspondence is found, it is possible to arrive at more realistic prediction values from simulation results. After carrying out non-parametric estimations of the probability distributions of the simulations, feature vectors are computed from these estimations, followed by a supervised learning that finds a mapping between the feature vectors issued from the simulations and the estimations of conditional probability distributions of the measurements. The proposed method is evaluated on a 3G radio network using indicators such as UpLink (UL) and DownLink (DL) base station loads. Results show that the proposed scheme is able to yield distributions that are much closer to measurements than simulations. With such a technique, it is possible to predict with enhanced accuracy new configurations and conditions for which we don't have observations.

This paper presents a use case scenario of IEEE P1900.4 standard designed to achieve temporal opportunistic spectrum access within IEEE802.11 bands. This offers a flexible framework that allows better utilization of radio resources and thus increases spectral efficiency. In this work, we propose a complete framework including scenario description with IEEE P1900.4 and ad-hoc reinforcement techniques for efficient opportunistic spectrum access. Thus, we propose to use the multi armed bandit approach at the terminal side. Results prove this method to achieve overall good results in performing opportunistic access within IEEE802.11 bands.

To enhance the generalization capacity of a distribution learning method, we propose to use a fuzzy Bayesian framework based on Bayes rules. The precision of the learning results is increased and the prediction quality is enhanced. The distribution learning method uses the information contained in the simulations and the knowledge of the measurements to learn a relation function. The fuzzy bayesian clustering (FBC) algorithm is a preprocessing technique that divides the whole learning space into subspaces where the generalization is better than the generalization into the whole space. We apply the FBC to a prediction tool of a third generation (3G) cellular radio network and results show that the generalization capacity is enhanced compared to classical clustering algorithms. Unobserved configuration can then be predicted with enhanced accuracy.

We have used measurements taken on real network to enhance the performance of our radio network planning tool. A distribution learning technique is adopted to realize this challenged task. To ensure better generalization capabilities of the learning algorithm, a preprocessing of data is required and involves the use of a clustering algorithm that divides the whole learning space into subspaces. In this paper we apply a new fuzzy clustering algorithm to a prediction tool of a third generation (3G) cellular radio network. Results show that the differences observed between simulations and measurements can be considerably diminished and the generalization capacity is enhanced thanks to the proposed clustering algorithm. This algorithm performs well than classical c-means algorithm. We can then predict with enhanced accuracy new configuration for which we don't have measurements, as long they are not very different from learned configurations.

This paper presents an efficient scheme to enhance the simulation results of a radio network planning tool by means of measurements. A multilayer perceptron (MLP) is trained to learn the mapping between the measurements and the simulations. The major contribution is the utilisation of independent component analysis (ICA) that transforms the inputs of the MLP into statistically independent variables and makes the complexity of MLP tractable. Other contributions consist of the use of the k-means clustering algorithm on the incoming data and the enrichment of the training data to enhance the generalization capability of the MLP. The proposed method is applied to a 3G mobile network to enhance the predictions of uplink (UL) and downlink (DL) base station loads. After a training performed on a given network configuration, mechanical antenna tilts are modified and we show that the results obtained by the supervised predictions are much closer to measurements than simulation results for cases that have not been encountered before

We propose a novel method to enhance the quality and precision of model-based simulation results by combining the a-priori information contained in the simulations with the a-posteriori knowledge of the measurements. This method involves the use of K-Means clustering, Independent Component Analysis (ICA) and Artificial Neural Network (ANN). The K-Means block divides the whole learning space into subspaces to ensure a better generalization of the ANN. The ICA block, being the main contribution of this work, makes the input variables of the ANN statistically independent so that the ANN can operate on one-dimensional distributions without losing information on joint statistics. The proposed method is applied to a prediction tool of a third generation (3G) cellular radio network. Results show that the differences observed between simulations and measurements can be considerably diminished. We can then predict with enhanced accuracy `unobserved' configurations as long they are not very different from `learned' configurations.

In this paper, we analyze the total ergodic spectral efficiency and sum-rates error-exponents of randomly-spread CDMA systems with diversity-combining linear multiuser receivers over time-varying frequency-selective multipath Rayleigh/Ricean fading channels. The basis of our analysis mainly being the joint/average eigenvalue density of the random cross-correlation matrices, our focus is on equal-rate/equal-energy identical users with statistically i.i.d. channels. The upper-bound ergodic spectral efficiency and the corresponding sum-rates error-exponent expressions are derived in closed-form in terms of the key system parameters for linear multiuser receivers with single-user ML decoding. The analysis sheds light onto asymptotic behaviour of channel-coded wideband CDMA systems over multipath fading channels.

We address in this paper the derivation and analysis of the outage spectral efficiencies achievable with linear multiuser receivers over randomly-spread multipath fading time-varying coded-CDMA channels. The basis of the derivations is the use of non-asymptotic average eigenvalue densities of random crosscorrelation matrices. The analyses give important clues on the achievable capacity with linear multiuser receivers under non-ergodic transmission situations.

A new unified method for closed- form theoretical performance analysis of randomly-spread CDMA systems over multi-path fading channels with multiuser receivers is presented. The basis of the analysis is the representation of the random signal-to-interference ratios (SIR) at finite system parameters in terms of the eigenvalues of the crosscorrelation matrices for which the non-asymptotic average distributions are found or known. The methodology presented complements the asymptotic limiting theory used in similar previous works [S. Verdu, S. Shamai, 1999, 2001 and R.R. Muller, 2000] over Gaussian/single-path fading channels, but at finite system parameters and over multipath fading channels. Though approach is non-asymptotic, it also validates asymptotically the predictions of the asymptotic limiting theory as the sizes of the system grow over all bounds. Application of the method is demonstrated by spectral efficiency derivations with linear multiuser receivers over multipath fading channels.

RAKE receivers suffer degradation in performance under fast fading due to errors in channel state estimation. The wide-sense stationary uncorrelated scatterer channel model tends itself to time-frequency canonical representation, and provides additional diversity in the frequency axis. That is, the joint multipath-Doppler diversity model enables the development of nonlinear decision-driven time-frequency multiuser detectors based on successive interference cancellation and multistage interference cancellation ideas. Simulation results demonstrate that the proposed nonlinear decision-driven time-frequency multiuser detectors promise substantially improved bit-error-rate performance compared to previously proposed time-frequency multiuser detectors while exhibiting strong resistance to the near-far effect.

We propose two pilot symbol aided iterative algorithms for channel re-estimation to be used in iterative detection and decoding (turbo-detection) of signals transmitted over frequency selective channels. The first algorithm is based on the expectation-maximization (EM) algorithm that makes use of a signal decomposition to compute separate dosed form recursion expressions for the channel coefficient re-estimates. The a posterior conditional channel state probabilities required by the EM-based turbo-estimation can be easily provided by the forward-backward trellis search performed by the soft-in soft-out (SISO) ISI decoding task. The second algorithm is based on the well-known bootstrap algorithm where soft information from the outer decoder is used to make hard decisions on transmitted coded data, and the detected coded data are used in the channel turbo-estimation process via the conventional linear pseudo-inverse method. Both of the proposed channel turbo-estimation algorithms are naturally embedded in the framework of the turbo-detector, bringing almost no overhead to the receiver complexity.

We investigate a communication model made of an outer code interleaved with a space-time trellis coded modulation (ST-TCM). Such a scheme is very relevant for highly frequency selective and time-varying channels using turbo-detection at the receiver side, but inherently suffers from a low coding rate. One possible solution to alleviate this impairment (i.e., to increase spectral efficiency) is to use parallel (or multilayered) transmission as done in recent approaches (BLAST). However, parallel transmission and highly frequency selective channels involve tremendous complexity at the receiver side. We propose a low complexity turbo-detector based on reduced-state trellis search and generalized per survivor processing that proves to be very efficient in this resulting multiple-input multiple-output (MIMO) ISI channel context.

RAKE receivers suffer degradation in performance under fast fading due to errors in channel state estimation. The wide-sense stationary uncorrelated scatterer channel model lends itself to time-frequency canonical representation, and provides additional diversity in the frequency axis. Using the joint spatial-multipath-Doppler diversity model based on this canonical representation, we propose a nonlinear space-time-frequency decorrelating decision-feedback multiuser detector for synchronous CDMA which utilizes decisions of the stronger users when forming the decisions for the weaker ones. Simulation results based on realistic fast-fading assumptions demonstrate that the proposed decorrelating decision-feedback multiuser detector promises substantially improved bit-error-rate Performance compared to space-time-frequency decorrelating multiuser detector for the weaker users, while the error probability of the weakest user approaches the single-user bound as the interferers grow stronger. The proposed detector exhibits strong resistance to the near-far effect for all users and the time-complexity-per-bit is linear in the number of users.

We analyze and compare several strategies for iteratively decoding trellis-encoded signals over channels with memory. Soft-in/soft-out extensions of reduced-complexity trellis search algorithms such as delayed decision-feedback sequence estimating (DDFSE) or parallel decision-feedback decoding (PDFD) algorithms are used instead of conventional BCJR and min-log-BCJR algorithms. It has been shown that for long channel impulse responses and/or high modulation orders where the BCJR algorithm becomes prohibitively complex, the proposed algorithms offer very good performance with low complexity. The problem of channel estimation in practical implementation of turbo detection schemes is studied in the second part. Two methods of channel reestimation are proposed: one based on the expectation-maximization (EM) algorithm and the second on a simple Bootstrap technique. Both algorithms are shown to dramatically improve the performance of the classical pseudo inverse channel estimation performed initially on a training sequence.

We investigate two iterative methods for decoding serially concatenated convolutional codes (SCCC) on multipath Rayleigh fading channels. One method employs optimal maximum a posterior (MAP) decoders or sub-optimum versions. The second slightly differs and is based on a soft-in soft-out (SISO) parallel decision-feedback decoder (PDFD), used for iteratively performing joint equalization and inner decoding together with outer decoding. Simulation results show that the first iterative process is only 0.7 dB better than the second one, which is however far less complex.

Code-hopping, which is an alternative spreading strategy to the static spreading code assignment of CDMA systems, is evaluated as a method to combat multiuser interference in S-CDMA systems. The uplink capacities of conventional and code-hopping S-CDMA are determined and compared by simulations carried out on a typical S-CDMA system for three nondispersive channels: no-fading, Rayleigh fading and Rician fading. The cellular structure is not composed of regular hexagons, but irregularly shaped natural cells where affiliation of the mobile user depends on the paths to the neighboring base stations. Results are presented for two different values of blocking ratio which is defined as the ratio of the number of rejected call attempts to the number of total call attempts, revealing significant uplink capacity increases provided by code-hopping in both fading and no-fading environments. Furthermore, it is observed that the increase in capacity is greater in fading channels than in no-fading ones.