ABSTRACT OF THE THESIS

Traffic offloading in HetNet using power biasing considering different path loss exponent

By

Jyotica Yadav

Master of Science in Electrical Engineering

University of California, Irvine, 2019

Professor Ender Ayanoglu, Chair

With the advent of new cellular technology, the number of users is expected to grow at a drastic rate that comes with the cost of an increase in data traffic. Therefore, to provide enough capacity for all the users, node density must be increased. One way to achieve that is through cell splitting but that would lead to higher chances of intercell interference along with incurring higher costs of installing high power macro nodes. Another way to attain a higher network capacity is to overlay a network of small low power nodes over an existing network thereby creating a heterogeneous network, i.e., HetNet [1]. In HetNet, the user is made available with more than one type of base station differing in the transmit power level and physical size. But, even with multiple options at hand, the user terminal always tends to associate itself to the macro base station i.e., the station with the highest transmit power. The reason is, that associativity is based on the received signal strength values i.e., RSS values, therefore, irrespective of the distance between the user and the station, the user will always have a preference towards macro station over low tier stations. Because of the above-mentioned claim, the overall reason behind shifting towards low tier station nullifies, therefore, it is essential to offload the data traffic from the macro tier. Some of the techniques to offload data traffic are offloading via power control, offloading via femtocell deployment and offloading via biasing, out of which, the latter is of concern to the thesis [2]. The research so far focuses on the stationary user and its connectivity to a base station. With user velocity in the picture, comes the concept of handover and its associated delays. Handover delays in dense networks lower the effect of an overall increase in network capacity as it includes the transmission of control signals rather than effective data transmission thereby raising the overhead over the network. Few handover skipping schemes have been proposed to avoid frequent handovers whose underlying assumption is same path loss exponents for all the tiers and unbiased power levels [3]. In this thesis, the effect of different path loss exponents and biased power levels has been investigated and the effect on coverage probability has been studied. In the end, the simulation results obtained are compared with the work done by other researchers so far.

The subject of B-ISDN came into being in the late 1980s, together with the concept of Asynchronous Transfer Mode (ATM). ATM is closely tied to high-speed packet switching by means of specialized switches implemented in hardware. Due to its high speed and packet structure, ATM technology was considered attractive to unify voice, data, and video services. A unification of these services over the telephone infrastructure was attempted earlier by a standards offering known as Integrated Services Digital Network (ISDN). Consequently, this new service unification was termed Broadband ISDN (B-ISDN). Although due to its origins, B-ISDN is sometimes closely tied to ATM technology, the term independently represents the vision of packet-based high-speed integration of voice, data, and video services. It is important that in this process, guarantees to satisfy different Quality-of-Service (QoS) needs (in terms of delay, loss, etc) required by voice, data, and video services are provided. In this vision, what is important is the unification, or integration of services; and the underlying technology is of secondary importance. As of the early 2000s, the technology to be employed in realizing this vision seems to have shifted from its origins of ATM. In this article, our emphasis is on B-ISDN as a service integration vision. Nevertheless, we will describe its original emphasis as the service offering of ATM as well as the path the industry seems to be taking in implementing this vision.

We present a new way of producing a channel chart in polar coordinates. We estimatethe angle of arrival θ and the distance between the base station and the user equipment ρ by employing our algorithms, inverse of the root sum squares of channel coefficients (ISQ) algorithm, linear regression (LR) algorithm, and the MUSIC/MUSIC (MM) algorithm. We compare these methods with the channel charting algorithms principal component analysis (PCA), Samson’s method (SM), and autoencoder (AE). We show that ISQ, LR, and MM outperform all three in performance. The performance of MM is much better than LR and ISQ in LOS channel conditions. MM has better performance than ISQ and LR but is more complex. ISQ and LR have similar performance with ISQ having less complexity than LR.

The number of connected wireless devices is anticipated to increase heavily in the next few years. Thus, there is a need for a new system which is able to handle billions of wireless devices. The category of massive multiple input multiple output (MIMO) systems is a great candidate for this purpose. Because of the large number of antennas in massive MIMO there is a need to reduce the dimension of the MIMO channel effectively to decrease the complexity. This could be achieved by using a particular prebeamforming technique that is introduced recently. An important aspect of wireless communication systems is the channel state information (CSI). In order to send and receive data through the channel, the transmitter and the receiver must know the CSI or at least have an estimation for it. In this thesis, different algorithms for estimating the channel vector coefficient and their performance are studied. Different approaches are used in order to find the best algorithm based on the performance of the estimating channel and the complexity of the algorithm. Also, algorithms are used to estimate the channel vector coefficient for different channel models.

The communication infrastructure is one of the biggest energy consumers in the world. With the expected exponential growth in the demand for wireless traffic, it becomes the foremost priority to make the communication systems energy-efficient. In this work, we will explore two energy efficiency enhancement techniques: Constant-Envelope modulation and Turbo coding. In Constant-envelope modulation, the high Peak-to-Average Power Ratio (PAPR) signal is transformed into a constant-envelope phase modulated signal. Thus the PAPR of the signal reduces to 0 dB, enabling the power amplifiers at the transmitter to work at the energy-efficient operational point. The second technique known as Turbo coding, has been known to perform very close to the theoretical bounds. Thus when Turbo codes are applied on a modulation scheme, there is a significant improvement in bit-error-rate performance. Consequently, the number of retransmissions is decreased which helps to conserve power at the transmitter.

In this thesis, we will explore the application of these two established techniques for the modulation schemes used in 3GPP LTE standards: Orthogonal Frequency Division Multiplexing (OFDM) and Single Carrier-Frequency Division Multiple Access (SC-FDMA). We will also present their comparisons in terms of bit-error-rate and spectral efficiency.

This work presents a study on single-carrier and orthogonal frequency-division multiplexing massive MIMO transmission systems for the frequency selective Gaussian multi-user downlink and uplink channels. It compares the spatially uncorrelated channels with the correlated ones in terms of the user achievable sum-rate. The correlation is considered to be among antenna elements at the base station (the transmitter side in the downlink direction and the receiver side in the uplink direction) in a channel with $K$ single-antenna users receiving and sending information from and to $M$ antennas at the base station. This work provides a general expression for the achievable rate by users in the channel and it shows for low-SNR regions in the channel, the channel matched filter technique outperforms all the other considered techniques, when the channel is uncorrelated over space. In a highly-correlated massive MIMO channel, conventional channel matched filter technique does not perform as expected. We show the failure of channel matched filter technique in the presence of two different correlation patterns among antennas at the base station and suggest better schemes which lead to a better achievable sum-rate. It has been shown as the result of computations that the suggested techniques (precoders in the downlink direction and equalizers in the uplink direction) perform better than the conventional channel matched filter technique in terms of achievable sum-rate.

The dissertation focuses on understanding parameter influences on variants of kernel regression models over graphs. Graphs are used to represent complex systems where components in the system are modeled as nodes, and relationships among the components are denoted as edges connecting nodes. Kernel regression models can be used to solve graph-related problems such as graph signal reconstruction and prediction. In the graph signal reconstruction problem, a common case is to predict an unknown attribute of a node using known values of the same attribute from other nodes and the network structure. In the graph prediction problem, a common case is to predict a graph signal over the network based on historical graph signals. The essence of the two problems is to model the input-output relationship, and the kernel-based regression model with an iterative solution is a simple yet possibly powerful solution. The dissertation will first show an application of the kernel regression model on the graph signal reconstruction problem over multi-layer graphs. The graph signal reconstruction problem aims to estimate unknown nodal values based on known nodal values and the multi-layer network structure. Viewing the mapping from the local network structure of a node to the nodal value as a function in a Reproducing Kernel Hilbert Space (RKHS), a regression model based on multiple kernels is built and a minimization problem is formatted. With the gradient descent algorithm, it is easy to find the solution to the minimization problem iteratively. In this application of the kernel-based regression model, the predicting ability of the model is verified. It is also seen from the application that the single-kernel models are used as building blocks of a multi-kernel model and that the performance is de- pendent on the hyper-parameter settings on the single-kernel regression models. To achieve better performance with less computational cost by selecting suitable hyper-parameters for the model, the dissertation then presents an analysis framework to analyze the influence of the hyper-parameters on the predictions of single-kernel regression models. Noting that due to the iterative nature of the model solution, it is hard to figure out the influence of the hyper-parameters directly. So, the main idea of the proposed framework is to express the model prediction as a weighted sum of the training observations, and then to analyze the influence of parameters on the observation weights. With the framework, it is found that the weights of the parameters are scaled kernel values of the input for prediction and inputs for training observations. This verifies that the kernel performs as a similarity measure and shows that the scaling factor for kernel values is related to the time difference between the two inputs of the kernel. The framework helps better understand the impact of the hyper parameters and hints at suitable selections of those parameters. After that, the framework is generalized to do parameter analysis for an iterative solution of a kernel regression model dealing with the graph signal prediction problem where the input is agnostic. In the generalized framework, the solution acquired from the batch gradient descent algorithm can be analyzed, making the solution acquired from the gradient descent algorithm a special case.

In recent years, there is a dramatic increase in the number of cellular devices. As a result, the

wireless network traffic increased as well. 4G, the most recent developed technology, soon

will not be good enough to fulfill the increasing need for capacity in the mobile data network.

As a consolidation of previous generation technologies, 5G is brought up as a solution. The

standardization for 5G is not established yet. Therefore, modulation techniques that can be

employed in 5G and which can support the need for the increased capacity are investigated.

In this thesis, modulation techniques CE-OFDM and CE-SC-FDMA are being investigated.

Their bit error rate performance is compared. It is found that both systems have close bit

error rate performance. A power amplifier is added to both systems and it is found that

power amplifier has no impact on the performance. Furthermore, CE-OFDMA and CE-

SC-FDMA are investigated and their bit error rate performance is found to be similar as

well.