Multivariate regression estimation of continuous-time processesfrom sampled data: local polynomial fitting approach

Let (Y,X)={Y(t),X(t),-∞j) be a renewal point processes on (0,∞), with a finite mean rate, independent of (Y,X). We consider the estimation of regression function r(x₀, x₁,...,x_m-1; τ₁,...,τ_m) of ψ(Y(τ_m)) given (X(0)=x₀, X(τ₁)=x₁,...,X(τ_m-1)=_x-1 ) for arbitrary lags 0<τ₁<...< τ_m on the basis of the discrete-time observations {Y(t_j),X(t_j),t_j)_j=1ⁿ . We estimate the regression function and all its partial derivatives up to a total order p⩾1 using high-order local polynomial fitting. We establish the weak consistency of such estimates along with rates of convergence. We also establish the joint asymptotic normality of the estimates for the regression function and all its partial derivatives up to a total order p⩾1 and provide explicit expressions for the bias and covariance matrix (of the asymptotically normal distribution)     

On ternary complementary sequences

A pair of real-valued sequences A=(a₁,a₂,...,a_N) and B=(b₁,b ₂,...,b_N) is called complementary if the sum R(·) of their autocorrelation functions R_A(·) and R_B(·) satisfies R(τ)=R_A(τ)+R _B(τ)=Σ_i=1^N$ -^τa_ia_i+τ+Σ_j=1 ^N-τb_jb_j+τ=0, ∀τ≠0. In this paper we introduce a new family of complementary pairs of sequences over the alphabet α₃=+{1,-1,0}. The inclusion of zero in the alphabet, which may correspond to a pause in transmission, leads both to a better understanding of the conventional binary case, where the alphabet is α₂={+1,-1}, and to new nontrivial constructions over the ternary alphabet α₃. For every length N, we derive restrictions on the location of the zero elements and on the form of the member sequences of the pair. We also derive a bound on the minimum number of zeros necessary for the existence of a complementary pair of length N over α₃. The bound is tight, as it is met by some of the proposed constructions, for infinitely many lengths     

In-circuit hot-carrier model and its application to inverter chainoptimization

An analytical in-circuit device lifetime model and design methodology for minimizing hot-carrier effects in an inverter chain are presented. Based on the model, in-circuit device lifetime, τ_AC , for hot-carrier induced degradation is given by τ_AC =τ_DC/R, R≡√π/2√Vcc/b√(T _rise.T_fall)/T_c where τ_DC is the device lifetime under a DC test, T_rise and T_fall are the rise time for input and fall time for output, T_C is cycle time and b is a constant. The model also shows that minimizing circuit delay in the inverter chain maximizes hot-carrier reliability     

On the cutoff point for pairwise enabling in multiple accesssystems

Let p⁰ be the minimum Bernoulli probability for which pairwise enabling is an optimal group testing algorithm under a Bernoulli arrival sequence model. In a previous work, it was shown that 0.430⩽p⁰⩽0.568 for unbounded Bernoulli arrival sequences, based on the threshold probabilities at which certain triple enabling algorithms (operating with and without the aid of a helpful genie, respectively) become more efficient. By deriving constructive results using the powerful but seemingly nonconstructive upper-bounding technique introduced by N.A. Mikhailov and B.S. Tsybakov (1981), the author sharpens this result by proving that p⁰ ⩽0.5 for unbounded arrival sequences, and that p⁰≈0.545 in the finite arrival sequence model recently studied by F.K. Hwang and X.M. Chang (1987). The present results for unbounded arrival sequences also extend to the reservation schemes considered by Hwang and Chang, where it is now shown that 0.386⩽p⁰_I⩽0.387 under the intermediate reservation model and 0.436⩽p⁰_G⩽1/√3 under the Gudjohnsen reservation model, respectively     

A new common-emitter hybrid-π small-signal equivalent circuitfor bipolar transistors with significant neutral base recombination

The linear superposition approach to the modeling of small-signal parameters in the presence of substantial base recombination, which involves a virtual transistor without base recombination, is identified to cause incorrect emitter current modeling. All of the terminal current changes can be correctly modeled by using the measured forced-V_BE Early voltage in a new equivalent circuit, which properly accounts for NBR and Early effect in a physically consistent manner. As a result, practical situations of small collector-base resistance (τ _μ) can be properly handled, τ_μ is related to the ac current-drive and ac voltage-drive Early voltages, which facilitates parameter extraction and circuit modeling. Measurements on state-of-the-art UHV/CVD SiGe HBT's show that the conventional assumption that τ_μ is far larger than the forced-V_BE output resistance τ₀ does not apply to devices with significant NBR. In practice, τ_μ can be comparable to (and smaller than) τ₀ depending on the device processing, profiles and operating temperature. Temperature dependent data are presented, and circuit implications are discussed based on the new equivalent circuit     

A model for the multipath delay profile of fixed wireless channels

This paper deals with the measurement and modeling of multipath delay on fixed wireless paths at 1.9 GHz in suburban environments. The primary focus is on the delay profile, which is the normalized plot of received power versus delay in response to an RT “impulse.” We describe measurement campaigns in the western suburbs of Chicago, IL, and in suburban north-central New Jersey. Our analysis of the data suggests to us that, for directive terminal antennas, the delay profile can be modeled as having a “spike-plus-exponential” shape, i.e., a strong return (“spike”) at the lowest delay, plus a set of returns whose mean powers decay exponentially with delay. This delay profile can be characterized by just two parameters (both variable over the terrain), namely, the ratio (K₀) of the average powers in the “spike” and “exponential” components and the decay time constant (τ₀) of the “exponential” component. No such simple structure appears to apply for delay profiles using omnidirectional antennas. For a directive antenna with a 32° beamwidth, we find that: (1) the statistical correlation between the profile parameters K₀ and τ₀ is negligible; (2) these parameters are relatively insensitive to antenna height and path length; and (3) over each measured region (Illinois and New Jersey), K₀ and τ₀ have median values close to 8 dB and just below 0.2 μs, respectively. Moreover, we have found simple probability distributions that accurately portray the variability of K₀ and τ₀ over the terrain     

Statistical model for the impulse response on infrared indoordiffuse channels

A statistical model for the IR-indoor-diffuse channel is presented, based on estimation of the parameters (τ_(rms) and τ_m) of the impulse response h(t). From these values, the authors compare the use of Rayleigh or Gamma distributions to fit the h(t) shape. A power budget approach is considered for estimating the gain power     

Effect of trap location and trap-assisted Auger recombination onsilicon solar cell performance

Model calculations were performed to investigate and quantify the effect of trap location and trap-assisted Auger recombination on silicon solar cell performance. Trap location has a significant influence on the lifetime behavior as a function of doping and injected carrier concentration in silicon. It Is shown in this paper that for a high quality silicon (τ=10 ms at 200 ohm-cm, no intentional doping), high resistivity (⩾200 ohm-cm) is optimum for high efficiency one sun solar cells if the lifetime limiting trap is located near midgap. However, if the trap is shallow (E_t-E_v⩽0.2 eV), the optimum resistivity shifts to about 0.2 ohm-cm. For a low quality silicon material or technology (10 μs at 200 ohm-cm, prior to intentional doping) the optimum base resistivity for one sun solar cells is found to be ~0.2 ohm-cm, regardless of the trap location. It is shown that the presence of a shallow trap can significantly degrade the performance of a concentrator cell fabricated on high-resistivity high-lifetime silicon material because of an undesirable injection level dependence in the carrier lifetime. The effect of trap assisted Auger recombination on the cell performance has also been modelled in this paper. It is found that the trap-assisted Auger recombination does not influence the one sun cell performance appreciably, but can degrade the concentrator cell performance if the trap-assisted Auger recombination coefficient value exceeds 2×10^-14 cm³/s. Therefore, it is necessary to know the starting lifetime as well as trap location in order to specify base resistivity in order to predict or achieve the best cell performance for a given one sun or concentrator cell design     

Statistical characterization of sample fourth-order cumulants of anoisy complex sinusoidal process

The paper deals with the statistical characterization of sample estimates of the fourth-order cumulants of a random process consisting of multiple complex sinusoids and additive colored Gaussian noise. In particular, it presents necessary and sufficient conditions for strong consistency of the sample cumulants of arbitrary orders, and derives expressions for the asymptotic covariance of the sample estimates of the fourth-order cumulants. It is shown that the fourth-order cumulant C_4y (τ₁,...,τ₄) can be written as a function of a single argument τ=τ₃+τ₄ -τ₁-τ₂, which implies large flexibility in estimating the cumulant. It is recommended that the estimate be based upon lags such that τ₁ is distant from τ₂ and τ₃ is distant from τ₄, and/or as a linear combination of such terms. The asymptotic variance of a cumulant-based frequency estimator is shown to have the form c₂·SNR^-2+c₃·SNR^-3 +c₄·SNR^-4, where the coefficient c ₂ may possibly vanish. The theory is illustrated via numerical examples. The results of this paper will be useful in analyzing the performance of various cumulant-based frequency estimation algorithms     

Effect of hot-electron injection of high-frequency characteristicsof abrupt In0.52(Ga1-xAlx)0.48As/InGaAs HBT's

The effect of hot-electron injection energy (E_i) into the base on the high-frequency characteristics of In_0/52(Ga_1-xAl_x)_0.48 As/InGaAs abrupt heterojunction bipolar transistors (HBTs) is investigated by changing the composition of the emitter. There exists an optimum E_i at which a maximum current gain cutoff frequency (ft) is obtained. Analysis of hot-electron transport in the base and collector by Monte Carlo simulation is carried out to understand the above phenomenon. The collector transit time (τ_c ) increases with E_i, because electrons with higher energy transfer from the Γ valley into the upper L and X valleys. At first, the base transit time (τ_b ) decreases with E_i at the low E_i region. However, τ_b does not decrease monotically with E_i, because of the nonparabolicity in the energy-band structure of InGaAs. Consequently, there exists a minimum in the sum of τ_b and τ_c , in other words a maximum f_t, at an intermediate value of E_i     

Laser-pumped rubidium frequency standards: new analysis andprogress

We have achieved a stability of 3·10^-13 τ ^-1/2 for 3<τ<30 s with a laser-pumped rubidium gas-cell frequency standard by reducing the effects due to noise in the microwave and laser sources. This result is one order of magnitude better than the best present performance of lamp-pumped devices     

Optimal covering polynomial sets correcting three errors for binarycyclic codes

The covering polynomial method is a generalization of error-trapping decoding and is a simple and effective way to decode cyclic codes. For cyclic codes of rate R<2/τ, covering polynomials of a single term suffice to correct up to τ errors, and minimal sets of covering polynomials are known for various such codes. In this article, the case of τ=3 and of binary cyclic codes of rate R⩾2/3 is investigated. Specifically, a closed-form specification is given for minimal covering polynomial sets for codes of rate 2/3⩽R<11/15 for all sufficiently large code length n; the resulting number of covering polynomials is, if R=2/3+ρ with ρ>0, equal to nρ+2V√nρ+(1/2) log_φ(n/ρ)+O(1), where φ=(1+√5)/2. For all codes correcting up to three errors, the number of covering polynomials is at least nρ+2√nρ+O(log n); covering polynomial sets achieving this bound (and thus within O(log n) of the minimum) are presented in closed-form specifications for rates in the range 11/15⩽R<3/4     

Delayed-self-heterodyne measurement of laser frequency fluctuations

The two-sample (or Allan) variance of frequency fluctuations was used for laser frequency-noise characterization at Fourier frequencies below 1 MHz. The delayed self-heterodyne technique was used for the variance measurement. Analytical calculations ensure that just twice the variance values were observed at τ<0.3τ_d, where τ is the averaging time for the frequency measurement and τ_d is the delay time. The variance measurements of 1.5-μm distributed Bragg reflector and distributed feedback laser diodes were demonstrated at 6.0×10^-7⩽τ⩽1.0×10^-4 [s] by using an 81.5-km fiber delay line. Observed flicker fluctuations (or 1/ f noise) are discussed     

Optimum optical power splitting ratio of decision drivenphase-locked loop in BPSK optical homodyne receiver

In a BPSK optical homodyne receiver that utilizes a decision-driven phase-locked loop, the splitting ratio of the received power and that of the local oscillator power are very important parameters in achieving high receiver sensitivity. This paper determines the optimum setting of these parameters considering the influence of the relative intensity noise of the local oscillator and the thermal noise of the preamplifier. The optimum splitting ratio of the local oscillator power to the Q-arm is found to be 0.5. The splitting ratio of the received power to Q-arm is obtained as a function of laser linewidth. The optimum setting of the received power and the local oscillator power Is independent of the relative intensity noise of the local oscillator, the thermal noise of the preamplifier and the bit rate, At the optimum splitting ratios, required beat linewidth is obtained as 1.3×10 ^-3/T_b(τ/T_b≪1) and 2.99×10 ^-3/τ(τ/T_b≫1), where T_b is the bit duration and τ is the loop propagation delay time. We show that the total power penalty of 0.8 dB from the shot noise limit can be realized with the relative intensity noise of -170 dB/Hz and equivalent input noise current of 10 pA/√(Hz), even if an imperfect balanced receiver is utilized; quantum efficiency ratio of the twin-photodetector is 0.96, propagation time difference T/T_b is 0.01. To confirm the theoretical model, a BPSK homodyne detection experiment is performed and good agreement is found between theoretical and experimental results     

The CEO problem [multiterminal source coding]

We consider a new problem in multiterminal source coding motivated by the following decentralized communication/estimation task. A firm's Chief Executive Officer (CEO) is interested in the data sequence {X(t)} _t=1^∞ which cannot be observed directly, perhaps because it represents tactical decisions by a competing firm. The CEO deploys a team of L agents who observe independently corrupted versions of {X(t)}_t=1^∞. Because {X(t)} is only one among many pressing matters to which the CEO must attend, the combined data rate at which the agents may communicate information about their observations to the CEO is limited to, say, R bits per second. If the agents were permitted to confer and pool their data, then in the limit as L→∞ they usually would be able to smooth out their independent observation noises entirely. Then they could use their R bits per second to provide the CEO with a representation of {X(t)} with fidelity D(R), where D(·) is the distortion-rate function of {X(t)}. In particular, with such data pooling D can be made arbitrarily small if R exceeds the entropy rate H of {X(t)}. Suppose, however, that the agents are not permitted to convene, Agent i having to send data based solely on his own noisy observations {Y_i(t)}. We show that then there does not exist a finite value of R for which even infinitely many agents can make D arbitrarily small. Furthermore, in this isolated-agents case we determine the asymptotic behavior of the minimal error frequency in the limit as L and then R tend to infinity     

Frequency-dependent characteristics of current distributions onmicrostrip lines

The spectral-domain analysis using Chebyshev's polynomials as basis functions is used to obtain the frequency-dependent characteristics of current distributions and the effective relative permittivities of an open microstrip line. The results obtained are compared to other available results. To obtain accurately the current distributions requires a larger number of basis functions. Both longitudinal and transverse current distributions on the strip are shown for wide ranges of frequency (0⩽h/λ₀⩽1)     

Insulator passivation of In0.2Ga0.8As-GaAssurface quantum wells

The electronic passivation of (100) In_0.2Ga_0.8 As-GaAs surface quantum wells (QWs) using in situ deposition of an amorphous, insulating Ga₂O₃ film has been investigated and compared to standard Al_0.45Ga_0.55As passivation. Nonradiative lifetimes τ_r=1.1±0.2 and 1.2±0.2 ns have been inferred from the dependence of the internal quantum efficiency η on optical excitation density P₀' for the Ga₂O₃ and Al_0.45Ga_0.55As passivated In_0.02 Ga_0.8As-GaAs surface QW, respectively. Beyond identical internal quantum efficiency, the amorphous Ga₂O₃ insulator passivation simplifies device processing, eludes problems arising from lattice-mismatched interfaces, and virtually eliminates band bending in electronic and optoelectronic devices based on a low dimensional system such as quantum wells, wires, and dots