Prapun Suksompong

Capacity Analysis for Integrate-and-Fire Neurons With Descending Action Potential Thresholds

Understanding how a biological neuron works has been a major goal in neuroscience. Under the Poisson-excitation assumption, results from earlier study by Suksompong and Berger on the timing jitter in the leaky integrate-and-fire (LIF) model of neurons are used to determine families of neural thresholding functions that are appropriate in certain interesting senses. Next, the neuron is treated as a communication channel for which information-theoretic quantities can be calculated. In particular, the optimal distribution of the Poisson excitation intensity is numerically evaluated along with the corresponding capacity using the Blahut-Arimoto algorithm. Simple formulas which approximate the optimal intensity distribution are given. Furthermore, the Jimbo-Kunisawa algorithm is used to explore energy-efficient operations for neuron. Finally, a rate-matching argument leads to a unique operating condition which turns out to agree with experimentally observed rate.

Enhancing PAPR performance of MIMO-OFDM systems using SLM technique with centering phase sequence matrix

Systems utilizing OFDM technique are plagued by the problem of high PAPR values in the transmitted signals. One popular solution is to apply selective mapping (SLM) technique. In this paper, SLM performance is studied in the context of MIMO-OFDM systems. The original SLM scheme for SISO-OFDM system is modified to use new phase sequence matrix based on a type of matrices call centering matrix. The new SLM scheme can be extended easily to the case of MIMO-OFDM systems. Its PAPR reduction performance is then compared with other modified SLM schemes. MATLAB simulations show that our proposed SLM modification significantly improves the PAPR reduction. It also outperforms some of the previously proposed SLM modifications.

A False Alarm Reduction Method for a Gas Sensor Based Electronic Nose

Electronic noses (E-Noses) are becoming popular for food and fruit quality assessment due to their robustness and repeated usability without fatigue, unlike human experts. An E-Nose equipped with classification algorithms and having open ended classification boundaries such as the k-nearest neighbor (k-NN), support vector machine (SVM), and multilayer perceptron neural network (MLPNN), are found to suffer from false classification errors of irrelevant odor data. To reduce false classification and misclassification errors, and to improve correct rejection performance; algorithms with a hyperspheric boundary, such as a radial basis function neural network (RBFNN) and generalized regression neural network (GRNN) with a Gaussian activation function in the hidden layer should be used. The simulation results presented in this paper show that GRNN has more correct classification efficiency and false alarm reduction capability compared to RBFNN. As the design of a GRNN and RBFNN is complex and expensive due to large numbers of neuron requirements, a simple hyperspheric classification method based on minimum, maximum, and mean (MMM) values of each class of the training dataset was presented. The MMM algorithm was simple and found to be fast and efficient in correctly classifying data of training classes, and correctly rejecting data of extraneous odors, and thereby reduced false alarms.

3-D localization algorithm using double-planar passive RFID arrays

RFID system uses sources emitting electromagnetic waves (RFID readers) to locate one or more targets (RFID tags), with the possible use of additional RFID tags as references. RFID positioning system is useful for localization, and is challenging for indoor environments. This research focuses on a 3-D based RFID positioning system, and the simulations are based on a hexahedron shape which can be a room or a container. The system enable many applications such as locating a small piece of object or person in a room or buildings, implementing indoor navigation system and so forth. Our scheme uses small number of readers whereas the locations are set to yield minimal interference. Small number of readers lead to lower cost of the system, and lower interference means higher performance for our system in a practical usage. Our calculation is based on the power level of each reader. In this paper, we propose the power searching scheme to improve the performance, and also a shift algorithm to improve the localization accuracy at the edge. With the help of a Gröbner basis, the estimation can be well accomplished.

Application of Pipelining Technique in Concatenated Tomlinson Harashima Precoder for Downlink MIMO Systems

We investigate the use of pipelining Tomlinson Harashima precoding (PIPTHP) for downlink multiple-input multiple-output (MIMO) systems. PIPTHP were previously analyzed in SISO systems but to our knowledge, implementation and analysis of PIPTHP in MIMO systems do not exist in the literature. Earlier studies have shown that PIPTHP technique can improve the BER performance and speed of the SISO systems. Here, the technique is extended to and analyzed in implementation for MIMO system. Furthermore, we develop and explore the idea of concatenated PIPTHP (C-PIPTHP) to improve the BER performance even more. The improved speed and bit error rate (BER) performance of C-PIPTHP are described in our simulation result. Our fast precoder can be implemented in 802.11 WLAN systems where MIMO has become the standard. We are also investigating an extension into multi-user (MU-) MIMO which is the technique of choice for the next generation of 802.11.

Jitter Analysis of Timing Codes for Neurons with Descending Action Potential Thresholds

We analyze three sources of timing jitter for (differential) time of arrival estimation for successive spikes arriving at a biological synapse. The dominant source of this jitter is believed to occur during spike generation. Other jitters introduced by propagation and estimation for time-of-arrival are on the order of 10 microseconds. We also obtain and compare several forms of the threshold curve when different assumptions about its form or error properties are imposed

Social-aware relay selection for device to device communications in cooperative cellular networks

Social information of users can be integrated in device to device communications to enhance performance of a cooperative cellular networks. This paper presents a Social-aware Midpoint Relay Selection Scheme which improves average throughput of the network by selecting a relay which is socially connected to the source and located near to midpoint of source and destination. We show the average throughput of our scheme can improve upon Pan and Wangs Hybrid Relay Selection Scheme (by upto 11%) as well as direct transmission.

Counter-intuitive channel allocation improvement in distributed cognitive radio networks by adding optimal interference in user's utility

Cognitive Radio Network (CRN) is a promising wireless communication solution whereby radio resources are automatically varied to optimize performance in the changing wireless environment. Our work focuses on distributed CRN with rational and independent users. The eventual state often reached in such a system is Nash Equilibrium (NE); multiple NEs with different performance exists. In this paper we segregate the data; individual users-throughput and interference received, at maximum total throughput and different grades of NE. Based on the characteristics of these metrics we develop policies for individual users which are implemented through their utility functions. Counter-intuitive policy is inferred and implemented; adding optimal amount of interference received by a user in its utility enhances the CRN performance measured in the form of normalized cumulative total throughput. At the optimal addition the CRN performance increases by 23% which is illustrated by the simulation results obtained from an average of 1,000 random scenarios.

Signal processing for multi-sensor E-nose system: Acquisition and classification

In this paper we review principle component analysis, linear discriminant analysis (LDA), A-nearest neighbor, feed forward backpropagation neural network, support vector machine, and radial basis function neural network (RBFNN) algorithms applied to electronic nose (E-Nose) for classification and detection. We show a method to extend the linear discriminant analysis (LDA) for multiclass (i.e. more than two class) LDA. By considering data alike typical E-Nose response we also show that RBFNN method need less time to classify new data. Thus RBFNN is more prominent in real time application for object identification from odor.

Near-optimality of the one-step max-sum-rate precoder with MMSE initialization for large downlink multi-user MIMO systems

Our recent study demonstrated that an iterative max-sum-rate (MSR) precoder with MMSE initialization gives an excellent sum-rate performance in downlink multi-user MIMO systems. As the number of transmit antennas increases, two crucial remarks were made: (1) its performance approaches the dirty-paper coding upper-bound and (2) the one-step iteration gives a decent performance. In this paper first, we further illustrate the second observation by looking at the optimality of this one-step max-sum-rate precoder with MMSE initialization (MSR-MMSE-1) for large MIMO systems. This shows that we can get near-optimal performance of the iterative MSR precoder without having to deal with any iteration at all. Then, we compare its sum-rate performance against one from the well-known block diagonalization algorithm (BD). Again, it is found that, when the number of transmit antennas is large enough, MSR-MMSE-1 outperforms BD across all SNR values under consideration.