The MicroWave Oven (MWO) is a popular appliance that acts as an unintentional interferer for 2.4 GHz IEEE 802.11 Wi-Fi communication signals. An interference mitigation technique is developed that incorporates cognitive radio paradigms allowing Wi-Fi devices to reliably transmit information while a residential MWO is operating. This technique is applied in the experimental case where Barker spread Wi-Fi signals carry data in the presence of MWO emissions. Bit error rate is evaluated to provide a performance metric for the mitigation technique.

The microwave oven (MWO) is a commonly available appliance that does not transmit data but still radiates signals in the unlicensed 2.4 GHz Industrial, Scientific and Medical (ISM) band. The MWO thus acts as an unintentional interferer for IEEE 802.11 Wireless Fidelity (Wi-Fi) communication signals. An analytic model of the MWO signal is developed and studied in this paper. The model's efficacy is studied via simulation and experimental emulation.

A study of interference caused by adjacent IEEE 802.11 Barker-code spread and complementary code keying spread wireless fidelity signals used in wireless local area networks is presented. The interference resulted from these signals is thoroughly analyzed and compared to that caused by bandlimited flat spectral interferers as well as narrow band interferers. Adaptive antennas employing tapped-delay-lines are used to mitigate the interference effect from these signals. The optimal values for the delay-line parameters to restore the performance at minimum cost are found

An effective method to match a desired group delay function using an all-pass digital filter is presented. This iterative eigen-delay technique uses the Rayleigh principle to solve the least squares delay error problem. Performance improvement over other design techniques is evident for low and medium order filters when a mean-squared error metric is computed.

A new domain, termed the frequency-delay is used to design stable, all-pass digital filters resembling a given delay response in the least-squares sense. This spectral technique identifies the delay response of a stable, second-order, all-pass digital filter as a double sideband suppressed carrier amplitude modulated signal in the frequency-delay domain. Iterative maximum likelihood techniques are used to render the filter coefficients. The algorithm is a significant improvement over related methods because it results in a physically realizable stable all-pass filter that closely approximates a desired delay response.

Current portable and land-mobile radio systems employ frequency modulation which exhibits a distinct capture effect. Furthermore, multipath propagation, shadowing, and the spatial distribution of the terminal units create data packets with varying signal strength. We investigate the performance of nonpersistent inhibit sense multiple access (ISMA) applied to the forward channel of packet radio communication networks. We show that the aggregate capacity of ISMA substantially increases compared to the noncapture case for small capture ratios. The mean packet delay of ISMA is also reduced with capture. The effect of different signal strength distribution is explored, and the capacity and delay improvements discussed.

The performance of a hybrid modulation technique derived from binary frequency shift keying is presented. In the hybrid scheme, each frequency is allowed to be phase modulated with any of p discrete, equally spaced phase shifts. The spectral separation between frequencies is chosen to make the carriers orthogonal, thus generating a multidimensional modulation scheme. The performance of complete and expurgated phase codes is examined for the Gaussian and Rayleigh fading channels, and their spectral characteristics are determined.>

The authors describe a new learning algorithm for the multi-layer perceptron. The learning problem is stated formally as an optimization problem that is solved through a series of systematic approximations. The solution uses the moments of the training data to design the network. This procedure has several advantages, most importantly the reduction in training time. The algorithm is verified and compared to backpropagation. In a speech recognition experiment the total training time was reduced by more than 75% when compared to backpropagation.>

Large neural networks can be segmented into several small subnetworks, trained independently, and combined. In the present work, a rule termed network fusion is derived for combining these subnetworks. This rule can be implemented with a single-layer neural network and is optimal if the output of the subnetworks is independent. The rule is demonstrated on a Gaussian random process and applied to a small vocabulary speech recognition system. The proposed method of combining the separately designed subnetworks into a large network exhibits a performance that approaches the Bayes limit.>

An iterative method for simultaneous pole-zero parameterization of a given filter impulse response is presented. An error criterion is constructed and iteratively minimized. The error criterion emphasizes both spectral peaks and spectral valleys in the desired signal spectrum and results in a set of linear equations that can be readily solved. A comparison with well-known methods (e.g. Shanks' and Kalman's methods) shows that this simultaneous iterative method is substantially better. Experimental results also show that a significant reduction in error is achieved after only three or four iterations. Applications of this method include system identification and pole-zero modeling of speech.>

An orthogonalized input-output characterization of nonlinear bandpass quadrature amplitude modulation systems with memory is developed. The orthogonal expansion is used with a nonlinear canceller structure to counteract the effect of nonlinear distortion introduced by high-power amplifiers at the transmitter. It is shown through simulations that when the transmitted symbols are known at the canceller, relatively simple structures can eliminate the effect of the nonlinearity. Even when the transmitted symbols are not known at the canceller, the improvement in performance with the nonlinear canceller, with respect to a linear receiver, is significant.>

The design and implementation of a parallel-processing-based pitch detector is presented. Pitch information is extracted by performing pitch detection on four different waveforms derived from the speech signal. Pitch information from the four pitch-detection processes is then combined to determine a final pitch estimate. The performance of this pitch detector is evaluated on a large database and compared to other well-known pitch detection algorithms. It has been implemented in real time on a TMS32020 fixed-point digital signal processor as part of a 2.4 kb/s vocoder. A performance comparison of the real-time fixed-point implementation and a computer simulation are also given. The results show that the pitch detector performance is maintained in the real-time implementation mainly because the majority of the algorithm computations are integer arithmetic and logic-type operations.>

One of the important characteristics of a high-definition television (HDTV) image is a wide aspect ratio. In some HDTV systems the image is segmented and different formats are used to transmit portions of the image. We address the problem of image reconstruction at the receiver and present an analysis of a wide aspect ratio HDTV system. Several image reconstruction techniques are described, including a channel-dependent design, and their effectiveness discussed. We also present a channel-independent system for image reconstruction that utilizes a vertical interval test line. Performance has been evaluated, via computer simulation, using still images. The results indicate an imperceptible amount of artifacts at the edge-center joints with an overlap as small as one picture element for the channel-independent reconstruction technique.

We present the results of a study to reduce the bit rate of speech that has been digitized with a continuously variable slope delta modulator (CVSD) operating at 16, 24, and 32 kbits/s. The theoretical reduction is found from the bit stream entropy. The actual reduction, via Huffman coding, is within 1-2 Percent of the theoretical value. The conditional entropy indicates that additional bit rate reduction can be achieved if we use a set of Huffman codes, conditioned on the past CVSD bits. A third technique, tandem coding, using a maximum likelihood predictor in tandem with run length and Huffman coding, is also investigated. Using these entropy techniques, bit rate reductions of 11-25 percent are achieved for the CVSD rates considered. The paper concludes with a study of the buffer requirements needed to support these entropy coders.

Many methods of speech waveform encoding utilize differential techniques for bandwidth compression. The channel bit rate of these systems is directly related to the slope characteristics of the speech signal. Employing well-known statistical characteristics of speech, this paper presents a derivation of the probability density function (pdf) for the differential slope of a speech signal and evaluates two families of pdfs. The technique used here, based on the Barrett-Lampard expansion, can also be applied to the general problem of obtaining the pdf of a random signal that has been digitally filtered. In this way, the frequency reliance upon a Gaussian assumption can be avoided.

This paper examines three new digital modulation binary codes developed by Hedeman and denoted as the H-1, H-2, and H-3 codes. It contains a complete description of the Hedeman codes and analyzes several of their important properties. The bit error probability is displayed as a function of signal-to-noise ratio, and detector implementation is discussed. It is shown that with a maximum likelihood sequence estimator using soft decisions, the H-2 and H-3 codes can achieve the same bit error rate as the nonreturn-to-zero (NRZ) or Manchester codes. Autocorrelation functions and power spectral densities of the Hedeman codes are presented. These results indicate that they possess the favorable quality of no dc component. Finally, symbol synchronization is considered. Here, phase synchronizing template patterns are identified and the probability of resynchronization is given as a function of time after loss of synchronization.

This paper constitutes a mathematical analysis of the effects of random channel errors when the transmitted signal is in adaptive delta modulation (ADM) format. This study not only permits us to model the behavior of an ADM signal, with respect to errors, but also provides a method for controlling excessive signal distortion, particularly for high bit error rates. Our analysis shows that the decoded error signal exhibits both a dc and an ac component. The dc error is controlled with a leaky accumulator and an appropriate output filter. The ac error is shown to be proportional to the original signal and we are able to control its amplitude by artificial error injection. This yields an effective way of improving performance in a noisy environment. We present computer simulation results to verify our analysis and error control with a variety of test signals.

This paper considers the realization of an all-digital system for direct filtering of a delta modulation (DM) encoded signal. We have shown [1] that this system can be realized fairly easily for any type of filter function when the signals are encoded with a linear DM. We now address the problem of digitally filtering an adaptive DM (ADM) encoded signal. We show that we can again realize any type of filter function and our system will operate directly on the ADM bit stream. In addition, we present a circuit realization that does not require complex digital hardware. Performance has been tested via simulation on a digital computer, using a variety of test signals; both temporal and spectral results are presented.

The purpose of this paper is to show that signals which are adaptive delta modulation (ADM) encoded can be arithmetically processed directly, without first decoding or converting to pulse code modulation (PCM) format. By operating on the serial DM bit streams, the sum, difference and product can be obtained in PCM and ADM format. Employing a four-term, non-recursive, averaging filter after the processors, we show that, for constant inputs, the signal-to-noise ratio (SNR) of the DM devices is exactly the same as that of their PCM counterparts. Although we have used the Song audio mode ADM [1] in the realization of our arithmetic processors, the designs are general enough to be applied to a large class of digital ADMs. To keep all systems practically realizable, we only employ operations which can be constructed with standard digital hardware, that is, adder, delays, hard-wired scalars and common logic circuits. Consequently, all our ADM devices can be manufactured with large scale integration where the distinct advantage is low cost and high reliability.