This paper proposes a lattice-reduction-aided MIMO-OFDM receiver with virtual channels; the receiver enables an increase in the downlink transmission speed for a user where the number of transmit antennas is considerably higher than that of the receive antennas. However, the receiver has a higher computational complexity than conventional lattice-reduction-aided MIMO receivers. Accordingly, we also propose novel techniques to reduce the computational complexity for the lattice-reduction-aided MIMO receivers with virtual channels. The proposed MIMO receiver achieves superior performance in 102 MIMO-OFDM systems. Furthermore, the proposed techniques are shown to reduce the computational complexity to approximately 40% of the original configuration in the 102 MIMO-OFDM systems.

This paper proposes a dynamic bandwidth allocation algorithm that improves the network performance and bandwidth sharing efficiency in the upstream channels of a hybrid passive optical network (PON) that combines a fiber-to-the-home (FTTH) access network and wireless sensor networks (WSNs). The algorithm is called the adaptive limited dynamic bandwidth allocation (ALDBA) algorithm. Unlike existing algorithms, the ALDBA algorithm is not limited to controlling just FTTH access networks, it also supports WSNs. For the proposed algorithm, we investigate the difference in the lengths of generated data packets between the FTTH terminals and sensor nodes of WSN to effectively evaluate the end-to-end average packet delay, bandwidth utilization, time jitter, and upstream efficiency. Two variants of the proposed algorithm and a limited service (LS) scheme, which is an existing well-known algorithm, are compared under non-uniform traffic conditions without taking into consideration priority scheduling. We demonstrate the proposed scheme through simulation by generating a realistic network traffic model, called self-similar network traffic. We conducted a detailed simulation using several performance parameters to validate the effectiveness of the proposed scheme. The results of the simulation showed that both ALDBA variants outperformed the existing LS scheme in terms of average packet delay, bandwidth utilization, jitter, and upstream efficiency for both low and high traffic loads.

In numerical simulations using massively parallel computers like GPGPU (General-Purpose computing on Graphics Processing Units), we often need to transfer computational results from external devices such as GPUs to the main memory or secondary storage of the host machine. Since size of the computation results is sometimes unacceptably large to hold them, it is desired that the data is compressed and stored. In addition, considering overheads for transferring data between the devices and host memories, it is preferable that the data is compressed in a part of parallel computation performed on the devices. Traditional compression methods for floating-point numbers do not always show good parallelism. In this paper, we propose a new compression method for massively-parallel simulations running on GPUs, in which we combine a few successive floating-point numbers and interleave them to improve compression efficiency. We also present numerical examples of compression ratio and throughput obtained from experimental implementations of the proposed method runnig on CPUs and GPUs.

Predicting network throughput is important for network-aware applications. Network throughput depends on a number of factors, and many throughput prediction methods have been proposed. However, many of these methods are suffering from the fact that a distribution of traffic fluctuation is unclear and the scale and the bandwidth of networks are rapidly increasing. Furthermore, virtual machines are used as platforms in many network research and services fields, and they can affect network measurement. A prediction method that uses pairs of differently sized connections has been proposed. This method, which we call connection pair, features a small probe transfer using the TCP that can be used to predict the throughput of a large data transfer. We focus on measurements, analyses, and modeling for precise prediction results. We first clarified that the actual throughput for the connection pair is non-linearly and monotonically changed with noise. Second, we built a previously proposed predictor using the same training data sets as for our proposed method, and it was unsuitable for considering the above characteristics. We propose a throughput prediction method based on the connection pair that uses ν-support vector regression and the polynomial kernel to deal with prediction models represented as a non-linear and continuous monotonic function. The prediction results of our method compared to those of the previous predictor are more accurate. Moreover, under an unstable network state, the drop in accuracy is also smaller than that of the previous predictor.

In order to reduce cost and improve efficiency, many data centers adopt virtualization solutions. The advent of virtualization allows multiple virtual machines hosted on a single physical server. However, this poses new challenges for resource management. Web workloads which are dominant in data centers are known to vary dynamically with time. In order to meet application's service level agreement (SLA), how to allocate resources for virtual machines has become an important challenge in virtualized server environments, especially when dealing with fluctuating workloads and complex server applications. User experience is an important manifestation of SLA and attracts more attention. In this paper, the SLA is defined by server-side response time. Traditional resource allocation based on resource utilization has some drawbacks. We argue that dynamic resource allocation directly based on real-time user experience is more reasonable and also has practical significance. To address the problem, we propose a system architecture that combines response time measurements and analysis of user experience for resource allocation. An optimization model is introduced to dynamically allocate the resources among virtual machines. When resources are insufficient, we provide service differentiation and firstly guarantee resource requirements of applications that have higher priorities. We evaluate our proposal using TPC-W and Webbench. The experimental results show that our system can judiciously allocate system resources. The system helps stabilize applications' user experience. It can reduce the mean deviation of user experience from desired targets.

In content services where people purchase and download large-volume contents, minimizing network traffic is crucial for the service provider and the network operator since they want to lower the cost charged for bandwidth and the cost for network infrastructure, respectively. Traffic localization is an effective way of reducing network traffic. Network traffic is localized when a client can obtain the requested content files from other a near-by altruistic client instead of the source servers. The concept of the peer-assisted content distribution network (CDN) can reduce the overall traffic with this mechanism and enable service providers to minimize traffic without deploying or borrowing distributed storage. To localize traffic effectively, content files that are likely to be requested by many clients should be cached locally. This paper presents a novel traffic engineering scheme for peer-assisted CDN models. Its key idea is to control the behavior of clients by using content-oriented incentive mechanism. This approach enables us to optimize traffic flows by letting altruistic clients download content files that are most likely contributed to localizing traffic among clients. In order to let altruistic clients request the desired files, we combine content files while keeping the price equal to the one for a single content. This paper presents a solution for optimizing the selection of content files to be combined so that cross traffic in a network is minimized. We also give a model for analyzing the upper-bound performance and the numerical results.

We propose a family of algorithms for efficiently merging on contemporary GPUs, so that each algorithm requires O(m log (+1)) element comparisons, where m and n are the sizes of the input sequences with m ≤ n. According to the lower bounds for merging all proposed algorithms are asymptotically optimal regarding the number of necessary comparisons. First we introduce a parallely structured algorithm that splits a merging problem of size 2l into 2i subproblems of size 2l-i, for some arbitrary i with (0 ≤ i ≤ l). This algorithm represents a merger for i=l but it is rather inefficient in this case. The efficiency is boosted by moving to a two stage approach where the splitting process stops at some predetermined level and transfers control to several parallely operating block-mergers. We formally prove the asymptotic optimality of the splitting process and show that for symmetrically sized inputs our approach delivers up to 4 times faster runtimes than the thrust::merge function that is part of the Thrust library. For assessing the value of our merging technique in the context of sorting we construct and evaluate a MergeSort on top of it. In the context of our benchmarking the resulting MergeSort clearly outperforms the MergeSort implementation provided by the Thrust library as well as Cederman's GPU optimized variant of QuickSort.

This paper presents our investigation into improving the performance of our previous automatic reading quality assessment system. The method of the baseline system is calculating the average value of the Phone Log-Posterior Probability (PLPP) of all phones in the voice to be assessed, and the average value is used as the reading quality assessment feature. In this paper, we presents three improvements. First, we cluster the triphones, and then calculate the average value of the normalized PLPP for each classification separately, and use this average values as the multi-dimensional assessment features instead of the original one-dimensional assessment feature. This method is simple but effective, which made the score difference of the machine scoring and manual scoring decrease by 30.2% relatively. Second, in order to assess the reading rhythm, we train Gaussian Mixture Models (GMM), which contain the information of each triphone's relative duration under standard pronunciation. Using the GMM, we can calculate the probability that the relative duration of each phone is conform to the standard pronunciation, and the average value of the probabilities is added to the assessment feature vector as a dimension of feature, which decreased the score difference between the machine scoring and manual scoring by 9.7% relatively. Third, we detect Filled Pauses (FP) by analyzing the formant curve, and then calculate the relative duration of FP, and add the relative duration of FP to the assessment feature vector as a dimension of feature. This method made the score difference between the machine scoring and manual scoring be further decreased by 10.2% relatively. Finally, when the feature vector extracted by the three methods are used together, the score difference between the machine scoring and manual scoring was decreased by 43.9% relatively compared to the baseline system.

In wireless sensor networks, unbalanced energy consumption and transmission collisions are two inherent problems and can significantly reduce network lifetime. This letter proposes an unequal clustering and TDMA-like scheduling mechanism (UCTSM) based on a gradient sinking model in wireless sensor networks. It integrates unequal clustering and TDMA-like transmission scheduling to balance the energy consumption among cluster heads and reduce transmission collisions. Simulation results show that UCTSM balances the energy consumption among the cluster heads, saves nodes' energy and so improves the network lifetime.

In this paper, we propose a novel incremental method for discovering latent variables from multivariate data with high efficiency. It integrates non-Gaussianity and an adaptive incremental model in an unsupervised way to extract informative features. Our proposed method discovers a small number of compact features from a very large number of features and can still achieve good predictive performance in EEG signals. The promising EEG signal classification results from our experiments prove that this approach can successfully extract important features. Our proposed method also has low memory requirements and computational costs.

In this paper, we reformulate the issue related to wireless mesh networks (WMNs) from the Chief Executive Officer (CEO) problem viewpoint, and provide a practical solution to a simple case of the problem. It is well known that the CEO problem is a theoretical basis for sensor networks. The problem investigated in this paper is described as follows: an originator broadcasts its binary information sequence to several forwarding nodes (relays) over Binary Symmetric Channels (BSC); the originator's information sequence suffers from independent random binary errors; at the forwarding nodes, they just further interleave, encode the received bit sequence, and then forward it, without making heavy efforts for correcting errors that may occur in the originator-relay links, to the final destination (FD) over Additive White Gaussian Noise (AWGN) channels. Hence, this strategy reduces the complexity of the relay significantly. A joint iterative decoding technique at the FD is proposed by utilizing the knowledge of the correlation due to the errors occurring in the link between the originator and forwarding nodes (referred to as intra-link). The bit-error-rate (BER) performances show that the originator's information can be reconstructed at the FD even by using a very simple coding scheme. We provide BER performance comparison between joint decoding and separate decoding strategies. The simulation results show that excellent performance can be achieved by the proposed system. Furthermore, extrinsic information transfer (EXIT) chart analysis is performed to investigate convergence property of the proposed technique, with the aim of, in part, optimizing the code rate at the originator.

A reliable detection of atrial fibrillation (AF) in Electrocardiogram (ECG) monitoring systems is significant for early treatment and health risk reduction. Various ECG mining and analysis studies have addressed a wide variety of clinical and technical issues. However, there is still room for improvement mostly in two areas. First, the morphological descriptors not only between different patients or patient clusters but also within the same patient are potentially changing. As a result, the model constructed using an old training data no longer needs to be adjusted in order to identify new concepts. Second, the number and types of ECG parameters necessary for detecting AF arrhythmia with high quality encounter a massive number of challenges in relation to computational effort and time consumption. We proposed a mixture technique that caters to these limitations. It includes an active learning method in conjunction with an ECG parameter customization technique to achieve a better AF arrhythmia detection in real-time applications. The performance of our proposed technique showed a sensitivity of 95.2%, a specificity of 99.6%, and an overall accuracy of 99.2%.

This paper presents an algorithmic approach to acquiring the influencing relationships among users by discovering implicit influencing group structure from smartphone usage. The method assumes that a time series of users' application downloads and activations can be represented by individual inter-personal influence factors. To achieve better predictive performance and also to avoid over-fitting, a latent feature model is employed. The method tries to extract the latent structures by monitoring cross validating predictive performances on approximated influence matrices with reduced ranks, which are generated based on an initial influence matrix obtained from a training set. The method adopts Nonnegative Matrix Factorization (NMF) to reduce the influence matrix dimension and thus to extract the latent features. To validate and demonstrate its ability, about 160 university students voluntarily participated in a mobile application usage monitoring experiment. An empirical study on real collected data reveals that the influencing structure consisted of six influencing groups with two types of mutual influence, i.e. intra-group influence and inter-group influence. The results also highlight the importance of sparseness control on NMF for discovering latent influencing groups. The obtained influencing structure provides better predictive performance than state-of-the-art collaborative filtering methods as well as conventional methods such as user-based collaborative filtering techniques and simple popularity.

Real-time on-chip measurement of bit error rate (BER) for high-speed analog-to-digital converters (ADCs) does not only require expensive multi-port high-speed data acquisition equipment but also enormous post-processing. This paper proposes a low-cost built-in-self-test (BIST) circuit for high-speed ADC BER test. Conventionally, the calculation of BER requires a high-speed adder. The presented method takes the advantages of Gray coding and only needs simple logic circuits for BER evaluation. The prototype of the BIST circuit is fabricated along with a 5-bit high-speed flash ADC in a 90-nm CMOS process. The active area is only 90 µm 70 µm and the average power consumption is around 0.3 mW at 700 MS/s. The measurement of the BIST circuit shows consistent results with the measurement by external data acquisition equipment.

A parameterization of perfect sequences over composition algebras over the real number field is presented. According to the proposed parameterization theorem, a perfect sequence can be represented as a sum of trigonometric functions and points on a unit sphere of the algebra. Because of the non-commutativity of the multiplication, there are two definitions of perfect sequences, but the equivalence of the definitions is easily shown using the theorem. A composition sequence of sequences is introduced. Despite the non-associativity, the proposed theorem reveals that the composition sequence from perfect sequences is perfect.

The expected write deficiency of the index-less indexed flash codes (ILIFC) is studied. ILIFC is a coding scheme for flash memory, and consists of two stages with different coding techniques. This study investigates the write deficiency of the first stage of ILIFC, and shows that omitting the second stage of ILIFC can be a practical option for realizing flash codes with good average performance. To discuss the expected write deficiency of ILIFC, a random walk model is introduced as a formalization of the behavior of ILIFC. Based on the random walk model, two different techniques are developed to estimate the expected write deficiency. One technique requires some computation, but gives very precise estimation of the write deficiency. The other technique gives a closed-form formula of the write deficiency under a certain asymptotic scenario.

The interference rejection combining (IRC) receiver, which can suppress inter-cell interference, is effective in improving the cell-edge user throughput. The IRC receiver is typically based on the minimum mean square error (MMSE) criteria, and requires a covariance matrix including the interference signals, in addition to a channel matrix from the serving cell. Therefore, in order to clarify the gain from the IRC receiver, the actual estimation error of these matrices should be taken into account. In a system performance evaluation, the link performance modeling of the IRC receiver, i.e., the output signal-to-interference-plus-noise power ratio (SINR) after IRC reception including the estimation errors, is very important in evaluating the actual performance of the IRC receiver in system level simulations. This is because these errors affect the suppression of the interference signals for the IRC receiver. Therefore, this paper investigates and proposes IRC receiver modeling schemes for the covariance matrix and channel estimation errors. As the modeling scheme for the covariance matrix, we propose a scheme that averages the conventional approximation using the complex Wishart distribution in the frequency domain to address issues that arise in a frequency selective fading channel. Furthermore, we propose a modeling scheme for the channel estimation error according to the ideal channel response of all cells and a channel estimation filter to address channel fading fluctuations. The results of simulations assuming the LTE/LTE-Advanced downlink with two transmitter and receiver antenna branches show that the proposed modeling scheme for the covariance matrix estimation error accurately approximates the performance of a realistic IRC receiver, which estimates the covariance matrix and channel matrix of the serving cell based on the demodulation reference signal (DM-RS), even in a frequency selective fading channel. The results also show that the proposed modeling scheme for the channel estimation error is a robust scheme in terms of the r.m.s. delay spread of a channel model compared to the scheme using the mean square error (MSE) statistic of the estimated channel coefficients based on a channel estimation filter.

In this paper, we have addressed the problem of ordering transactions in network-on-chips (NoCs) for post-silicon validation. The main idea is to extract the order of the transactions from the local partial orders in each NoC tile based on a set of “happened-before” rules, assuming transactions do not have a timestamp. The assumption is based on the fact that implementation and usage of a global time as timestamp in such systems may not be practical or efficient. When a new transaction is received in a tile, we send special messages to the neighboring tiles to inform them regarding the new transaction. The process of sending those special messages continues recursively in all the tiles that receive them until another such special message is detected. This way, we relate local orders of different tiles with each other. We show that our method can reconstruct the correct transaction orders when communication delays are deterministic. We have shown the effectiveness of our method by correctly ordering the transaction in NoCs with mesh and torus topologies with different sizes from 5*5 to 9*9. Also, we have implemented the proposed method in hardware to show its feasibility.

This paper investigates the dominant impact on the interference rejection combining (IRC) receiver due to the downlink reference signal (RS) based covariance matrix estimation scheme. When the transmission modes using the cell-specific RS (CRS) in LTE/LTE-Advanced are assumed, the property of the non-precoded CRS is different from that of the data signals. This difference poses two problems to the IRC receiver. First, it results in different levels of accuracy for the RS based covariance matrix estimation. Second, assuming the case where the CRS from the interfering cell collides with the desired data signals of the serving cell, the IRC receiver cannot perfectly suppress this CRS interference. The results of simulations assuming two transmitter and receiver antenna branches show that the impact of the CRS-to-CRS collision among cells is greater than that for the CRS interference on the desired data signals especially in closed-loop multiple-input multiple-output (MIMO) systems, from the viewpoint of the output signal-to-interference-plus-noise power ratio (SINR). However, the IRC receiver improves the user throughput by more than 20% compared to the conventional maximal ratio combining (MRC) receiver under the simulation assumptions made in this paper even when the CRS-to-CRS collision is assumed. Furthermore, the results verify the observations made in regard to the impact of inter-cell interference of the CRS for various average received signal-to-noise power ratio (SNR) and signal-to-interference power ratio (SIR) environments.

Nano-scale VLSI design is facing the problems of increased test data volume. Small delay defects are becoming possible sources of test escapes, and high delay test quality and therefore a greater volume of test data are required. The increased test data volume requires more tester memory and test application time, and both result in test cost inflation. Test pattern ordering gives a practical solution to reduce test cost, where test patterns are ordered so that more defects can be detected as early as possible. In this paper, we propose a test pattern ordering method based on SDQL (Statistical Delay Quality Level), which is a measure of delay test quality considering small delay defects. Our proposed method orders test patterns so that SDQL shrinks fast, which means more delay defects can be detected as early as possible. The proposed method efficiently orders test patterns with minimal usage of time-consuming timing-aware fault simulation. Experimental results demonstrate that our method can obtain test pattern ordering within a reasonable time, and also suggest how to prepare test sets suitable as inputs of test pattern ordering.