| 1. |
|
|
| 2. |
|
|
| 3. |
- Sjödin, Martin, 1974-, et al.
(författare)
-
Sustainable Batteries
- 2014
-
Ingår i: NFM conference, Prague 16-18th June 2014..
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)
|
|
| 4. |
|
|
| 5. |
|
|
| 6. |
|
|
| 7. |
- Huang, Jin
(författare)
-
Sequential Data Learning, Scalable Models and Adversarial Regularization
- 2023
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Time Series Prediction (TSP) has been used in mobile network traffic data analysis to produce predictive results for network planning and resource allocation. In the first part of this thesis, we propose a novel method of predicting mobile network traffic using neural networks based on conditional probability modeling between adjacent data windows in the time series sequence. Firstly, we develop a pre-processing method to aggregate the raw traffic log data and sample the aggregated time series to adjacent data windows, as training samples. Secondly, we use neural networks to parameterize the conditional probability between adjacent data windows and estimate the probability by training the neural networks with sampled data. The estimated conditional probability is then used to ensemble the prediction. Thirdly, we show theoretically that the prediction based on all historical data is equivalent to the prediction based on just previous data window, given the estimation of conditional probability between adjacent data windows. We also analyze computation complexity and show that seasonality will reduce the computational complexity. In the experiment, we compare the prediction performance among the models with different seasonality, sample size and number of hidden layers, and show that the proposed schemes achieve better prediction accuracy than state-of-the-art.The Recurrent Neural Networks (RNN) with richly distributed internal states and flexible non-linear transition functions, havegradually overtaken the dynamic Bayesian networks in modeling highly structured sequential data. These data, which may come fromspeech and handwriting, often contain complex relationships between the underlying variational factors such as speakercharacteristic and the observed data. The standard RNN model has very limited randomness or variability in its structure, which comes from the output conditional probability model. To improve the variability and performance, we study the new latent variable models with novel regularization methods. The second part of this thesis will present different ways of using high level latent random variables in RNN to model the variability in the sequential data. We will explore possible ways of using adversarial methods to train a variational RNN model. Through theoretical analysis we show that, contrary to competing approaches our schemes have theoretical optimum in the model training and the symmetric objective function in the adversarial training provides better model training stability. Our approach also improves the posterior approximation in the variational inference network by a separated adversarial training step. Numerical results simulated from TIMIT speech data show that reconstruction loss and evidence lower bound converge to the same level and adversarial training loss converges in a stable course. The results also show our approach of regularization provides stability and smoothness on probability distribution distance minimization between prior and posterior of the latent variables. In the last part of this thesis, we studies potential challenges and opportunities in intelligent road traffic sensing from the data mining and learning point of view with mobile network generated data. This part of the thesis only include qualitative analysis. Firstly, we classify the data resources available in the commercial mobile network according to different taxonomy criteria. Then, we outline the broken-down problems that fit in the framework of road traffic sensing based on mobile user network log data. We study the existing data processing and learning algorithms on extracting road traffic condition information from a large amount of mobile network log data. Finally we make suggestion on potential future work for road traffic sensing on data from mobile networks.
|
|
| 8. |
- Huang, Shaocheng, 1990-
(författare)
-
Rate, Reliability and Secrecy Performance Analysis and Optimization for Millimeter WaveCommunications
- 2020
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- With the fast development of electronic devices and computer science, various emerging applications (e.g., virtual and augmented reality, ultra-high-definition three-dimensional video, autonomous driving, big data analysis, etc.) have created an explosive growth of mobile data traffic and caused growing demands for higher communication rates, more reliable and secure connectivity in the future wireless communications, e.g., the fifth-generation (5G) and beyond mobile communications. In recent years, millimeter wave (mmWave) communication, as a promising candidate to meet the aforementioned demands, has attracted extensive research attention, and is regarded as one of the key enablers for the 5G and beyond mobile communications. The main features of mmWave communications include: abundant spectral resources, high penetration loss, severe pathloss, and narrow antenna beams, and these particular features make the potential challenges and solutions with mmWave significantly different from those in the conventional sub-6 GHz systems. It is known that beamforming is a crucial stage of mmWave communication to support high antenna gains and suppress inter-channel interference. However, the related research on beamforming design is fairly recent and insufficient. Motivated by the urgent needs for further development, in this thesis, we investigate beamforming optimization and the reliability and secrecy performance for mmWave communications. Our main research regarding beamforming design and secrecy performance can be categorized into the following three aspects: 1) Hybrid beamforming (HBF) for mmWave systems with learning machines: We propose two low-complexity and robust learning schemes to design HBF for mmWave full-duplex systems and multi-user multi-input and multi-output (MU-MIMO) systems i.e., extreme learning machine based HBF and convolutional neural networks based HBF. To provide accurate labels for off-line training, effective HBF algorithms are proposed to achieve joint self-interference cancellation and HBF optimization for mmWave full-duplex systems and to achieve joint transmitting and receiving HBF optimization for mmWave MU-MIMO systems. The convergence of the proposed algorithms is proven and the computation complexity is analyzed. Results show that the proposed learning based methods can achieve higher spectral efficiency, lower complexity, and more robust HBF performance than conventional optimization based methods. 2) Decentralized beamforming for intelligent reflecting surface (IRS)-enhanced cell-free networks: To avoid the centralized computation and inspired by the cost-effective IRS technique, we propose a fully decentralized design framework for cooperative beamforming in IRS-aided cell-free networks, in which transmitting digital beamformers and IRS-based analog beamformers are jointly optimized. We first derive a closed-form expression of each updating variable and then propose a fully decentralized beamforming scheme based on the alternating direction method of multipliers to incrementally and locally update the beamformers. Results reveal that the new scheme outperforms existing decentralized methods and IRSs can significantly improve the system sum-rate. 3) Physical layer security of mmWave non-orthogonal multiple access (NOMA) networks: Considering the limited scattering characteristics of mmWave channels and imperfect successive interference cancellation at receivers, we develop an analytic framework on the secrecy outage probability for mmWave NOMA networks. Based on the directional transmission property of mmWave signals, we propose a minimal angle-difference user pairing scheme to reduce the secrecy outage probability of users. Considering the spatial correlation between the selected user pair and eavesdroppers, we develop two maximum ratio transmission beamforming schemes to further enhance the secrecy performance of mmWave NOMA networks. Closed-form secrecy outage probability for the paired users with different eavesdropper detection capacities is derived 4) Achievable rates and reliability analysis of mmWave channels: We leverage random coding error exponent to investigate the achievable rate of mmWave channels under reliability and packet duration (finite blocklength) constraints. Under the assumption of perfect and imperfect channel state information at the receiver (CSIR), exact and approximate analytical expressions of achievable rates are derived to capture the relationship of rate, latency, and reliability. Furthermore, we show that the achievable rate always increases as the bandwidth increases with perfect CSIR. However, there exists a critical bandwidth that maximizes the achievable rate for non-line-of-sight mmWave signals with imperfect CSIR, beyond which the achievable rate will decrease with increasing bandwidth. For imperfect CSIR, the optimal training symbol length and power allocation factor at the training phase are investigated and closed-form expressions for special cases are derived.
|
|
| 9. |
|
|
| 10. |
- Huang, Xiao, 1987-
(författare)
-
Conducting Redox Polymers for Electrode Materials : Synthetic Strategies and Electrochemical Properties
- 2017
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Organic electrode materials represent an intriguing alternative to their inorganic counterparts due to their sustainable and environmental-friendly properties. Their plastic character allows for the realization of light-weight, versatile and disposable devices for energy storage. Conducting redox polymers (CRPs) are one type of the organic electrode materials involved, which consist of a π-conjugated polymer backbone and covalently attached redox units, the so-called pendant. The polymer backbone can provide conductivity while it is oxidized or reduced (i. e., p- or n-doped) and the concurrent redox chemistry of the pendant provides charge capacity. The combination of these two components enables CRPs to provide both high charge capacity and high power capability. This dyad polymeric framework provides a solution to the two main problems associated with organic electrode materials based on small molecules: the dissolution of the active material in the electrolyte, and the sluggish charge transport within the material. This thesis introduces a general synthetic strategy to obtain the monomeric CRPs building blocks, followed by electrochemical polymerization to afford the active CRPs material. The choice of pendant and of polymer backbone depends on the potential match between these two components, i.e. the redox reaction of the pendant and the doping of backbone occurring within the same potential region. In the thesis, terephthalate and polythiophene were selected as the pendant and polymer backbone respectively, to get access to low potential CRPs. It was found that the presence of a non-conjugated linker between polymer backbone and pendant is essential for the polymerizability of the monomers as well as for the preservation of individual redox activities. The resulting CRPs exhibited fast charge transport within the polymer film and low activation barriers for charge propagation. These low potential CRPs were designed as the anode materials for energy storage applications. The combination of redox active pendant as charge carrier and a conductive polymer backbone reveals new insights into the requirements of organic matter based electrical energy storage materials.
|
|
