Bruce F. McGuffin

An efficient algorithm for two-dimensional autoregressive spectrum estimation

An algorithm similar to the Burg algorithm is given for autoregressive estimation in one dimension. Coefficients of the filter used to decorrelate the signal are found recursively from lower-order filters. The coefficients needed to get the larger-sized filters can be found at each stage of the recursion from the correlations of lower-order filter outputs. Thus, the resulting estimate does not assume knowledge of ensemble averages. It offers a computational saving of 25% over the commonly used method, in which an estimate of the correlation matrix is first computed and then inverted. >

Jammed FH-FSK performance in Rayleigh and Nakagami-m fading

This paper looks at the performance of frequency hopped spread spectrum (FH) binary frequency shift keyed (FSK) modulated signals in wideband and partial-band noise jamming, when either the signal or signal and jammer are subject to Rayleigh or Nakagami-m fading. Simple bit error rate (BER) approximations are derived as functions of the signal and jammer fading parameters (which need not be equal) using the assumption that average jammer power is much greater than average signal power. The approximations are used to find the worst case jammer fractional bandwidth. For signals with Rayleigh or more severe fading, full-band jamming is the optimal attack. In less severe signal fading, partial-band jamming may be more disruptive, depending on the available jammer power. It is found that jammer fading improves communications channel performance with wideband jamming by 2-4 dB effective power gain. When jammer bandwidth is optimized, BER performance is dominated by the signal fading parameter, although the optimal jammer bandwidth will vary somewhat due to jammer fading. The accuracy of these approximations is verified by comparing them to more detailed numerical solutions.

The effect of bandpass limiting on a signal in wideband noise

The authors present a way to find the change in signal-to-noise-density ratio when a constant amplitude signal embedded in wideband noise passes through a bandpass limiter. This ratio is a more accurate predictor of demodulation or detection performance than the commonly used signal-to-noise-power ratio of the limiter output if signal processing includes narrowband elements such as matched filters. The noise bandwidth out of the bandpass limit is at least as wide as the input noise bandwidth, and subsequent signal processing will remove much of the output noise before a decision is made. The present result is given in the form of an infinite series for the limiter output noise power spectrum density at the signal frequency. This can be combined with a known expression for output signal power to find the change in signal-to-noise-density ratio.<<ETX>>

Distributed jammer performance in Rayleigh fading

A distributed jammer is a set of small jamming transmitters distributed over a region to disrupt tactical communications. Propagation losses experienced by terrestrial jammers will be comparable to losses seen by the target radio links, with multipath fading, and power failing off faster than 1/r/sup 2/. An intelligent distributed jammer can partly overcome excess propagation loss by exploiting the proximity and small number of targets in each nodes coverage region to extract information needed for frequency following or synchronized pulsed jamming. The coverage radius of an individual jammer node is an important system parameter, determining the number of nodes required. and the resulting jammer cost. A set of equations are presented that give the probability of jamming a target receiver randomly located inside a fixed range, when both communications and jamming signals suffer from Rayleigh fading. and average jammer power falls off faster than line-of-sight power. An upper bound on jammer performance improvement with intelligent jamming is found by assuming the nodes are able to extract desired target signal information. and calculating the improvement in jammer reliability with a fixed coverage range. Both frequency followers and synchronized pulsed jammers are examined. Performance improvement is greatest when each nodes coverage region contains a limited number of distinct communications sub-networks.

High resolution frequency hopped symmetric DPSK in tone jamming

Symmetric DPSK (SDPSK) is a modified form of DPSK, with data values coded as /spl plusmn/90/spl deg/ phase shifts. This modification reduces spectrum regrowth of filtered signals in nonlinear amplifiers, and sensitivity to intersymbol interference caused by filtering. On an ideal channel, SDPSK performance in optimized partial band noise (PBNJ) jamming is the same as DPSK performance under the same conditions. In traditional frequency hopping (FH) the signal jumps among a set of allowed carrier frequencies separated by the signal bandwidth. To improve resistance to tone jammers, high resolution frequency hopping (HRFH) uses many more allowed frequencies, much closer together. The probability of the signal hitting a tone does not change, but the jammer cannot control tone location relative to the signal. The authors investigate HRFH SDPSK performance in tone jamming. For easy of analysis, the effects of channel filtering and AWGN are ignored. Jamming tones have equal power, and inter-tone spacing is no less than the signal bandwidth. Communication performance is found and the tone jamming strategy optimized for this model. It is shown that multitone jamming requires 0.5 dB less power than PBNJ to effect the same degradation. Given the simplifications used, this difference is not significant.<<ETX>>

DCS-1900 system configuration

DCS-1900 is a proposed PCS system based on the GSM mobile telephone system. Like GSM, many configuration options and parameters must be selected by the base station manufacturer, or system operator. This paper examines parameter effects on system performance, to determine good configuration strategies. Issues considered include frequency plan, power control, base station synchronization, and DTX. A lower bound is presented for DCS-1900 uplink performance, and is used to determine the relationship between capacity and coverage. The best frequency plan is found to be a frequency hopped system with complete frequency reuse in every cell. Collisions between signals are controlled by restricting cell loading. The 1/3 frequency plan is shown to have at least 99% coverage with 0.24 loading, supporting 144 calls per cell in 30 MHz bandwidth.

High resolution frequency hopped DPSK in tone jamming

An analysis for the bit error ratio (BER) of high resolution frequency-hopped differential phase shift keyed (DPSK) signals in multitone jamming is presented. Tones are assumed to have equal power and may be either uniformly spaced or at random frequencies. Under the constraint that the probability of two tones hitting a signal at once is small, the degradation due to this jamming scenario is shown to be no worse than that of optimized partial band noise jamming.<<ETX>>

Mobile MIMO capacity in network interference: Informed and Uninformed Transmitters

Multiple-Input Multiple-Output (MIMO) receivers maximize link throughput by finding the best balance between enhancing signal power and suppressing interference. This makes MIMO a good candidate for Mobile Ad-hoc Networks (MANETs) with network interference. It has been shown previously that MIMO achieves higher throughput in network interference if transmitters reduce the number of transmit antennas or modes (data streams) used. That result is confirmed here using a more realistic model than that previously employed, including Rice fading, log-normal shadowing, and range-loss exponents and shadowing standard deviations that grow with range. Ordinarily, limiting the number of transmit modes requires some sort of enforcement policy. Informed Transmitter (IT-) MIMO links with Channel State Information (CSI) feedback optimize single-link performance in severe interference by reducing the number of transmit modes and concentrating available power in the strongest. It is shown here that this reduction in IT-MIMO signal complexity results in network performance close to that of network-optimized Uninformed Transmitter (UT-) MIMO, without the need for network coordination.

Moderate time-scale dynamic spectrum access with wide-area spectrum sensors

Simulation approaches and results are provided for Dynamic Spectrum Access (DSA) with remote wide-area spectrum monitoring equipment and moderate delay in spectrum data dissemination. Issues expected to arise using this approach include information delay time, failure to detect all transmitters, transmitter location errors, poor transmit-power estimates, and errors in the propagation model used to predict radio interference. The system works well if delays do not exceed 10 minutes, location error standard deviations are less than or equal to 2 km, and at least half of the transmitters are detected. It is found to be more sensitive to poor transmitter power and propagation-loss estimates.

LTPSK: simple, bandwidth efficient modulation for multi-rate satellite channels

A constant envelope modulation called linear-transition phase shift keying (LTPSK) is examined as a candidate for use as a bandwidth and power efficient modulation on nonlinear satellite channels. This modulation is similar to differential phase shift keying (DPSK), but with phase transitions realized as a linear phase ramp between symbols to decrease spectral occupancy without introducing amplitude variation. The ratio of phase-transition time to symbol duration is referred to as the LTPSK mode. Spectral occupancy and BER performance with differential detection vary between that of DPSK and differential minimum shift keying (MSK), depending on the mode selected. By making the LTPSK mode proportional to the data rate, a simple LTPSK modulator and differential demodulator can support multiple data rates without the need for a bank of filters. The highest rate supported has the lowest ratio of spectral occupancy to data rate, and spectral occupancy is approximately constant for all rates, allowing efficient spectrum allocation with fixed frequency assignments. >