arXiv subjects
Publications and source records attributed to Amos Lapidoth.
Ahlswede, Cai, and Zhang proved that, in the noise-free limit, the zero-undetected-error capacity is lower bounded by the Sperner capacity of the channel graph, and they conjectured equality. Here we derive an upper bound that proves the conjecture.
A task is randomly drawn from a finite set of tasks and is described using a fixed number of bits. All the tasks that share its description must be performed. Upper and lower bounds on the minimum $ρ$-th moment of the number of performed tasks are derived. The key is an analog of the Kraft Inequality for partitions of finite sets. When a sequence of tasks is produced by a source of a given Rényi entropy rate of order $1/(1+ρ)$ and $n$ tasks are jointly described using $nR$ bits, it is shown that for $R$ larger than the Rényi entropy rate, the $ρ$-th moment of the ratio of performed tasks to $n$ can be driven to one as $n$ tends to infinity, and that for $R$ less than the Rényi entropy rate it tends to infinity. This generalizes a recent result for IID sources by the same authors. A mismatched version of the direct part is also considered, where the code is designed according to the wrong law. The penalty incurred by the mismatch can be expressed in terms of a divergence measure that was shown by Sundaresan to play a similar role in the Massey-Arikan guessing problem.
A coding scheme is proposed for the memoryless Gaussian broadcast channel with correlated noises and feedback. For all noise correlations other than -1, the gap between the sum-rate the scheme achieves and the full-cooperation bound vanishes as the signal-to-noise ratio tends to infinity. When the correlation coefficient is -1, the gains afforded by feedback are unbounded and the prelog is doubled. When the correlation coefficient is +1 we demonstrate a dichotomy: If the noise variances are equal, then feedback is useless, and otherwise, feedback affords unbounded rate gains and doubles the prelog. The unbounded feedback gains, however, require perfect (noiseless) feedback. When the feedback links are noisy the feedback gains are bounded, unless the feedback noise decays to zero sufficiently fast with the signal-to-noise ratio. Extensions to more receivers are also discussed as is the memoryless Gaussian interference channel with feedback.
We study an interference network where equally-numbered transmitters and receivers lie on two parallel lines, each transmitter opposite its intended receiver. We consider two short-range interference models: the "asymmetric network," where the signal sent by each transmitter is interfered only by the signal sent by its left neighbor (if present), and a "symmetric network," where it is interfered by both its left and its right neighbors. Each transmitter is cognizant of its own message, the messages of the $t_\ell$ transmitters to its left, and the messages of the $t_r$ transmitters to its right. Each receiver decodes its message based on the signals received at its own antenna, at the $r_\ell$ receive antennas to its left, and the $r_r$ receive antennas to its right. For such networks we provide upper and lower bounds on the multiplexing gain, i.e., on the high-SNR asymptotic logarithmic growth of the sum-rate capacity. In some cases our bounds meet, e.g., for the asymmetric network. Our results exhibit an equivalence between the transmitter side-information parameters $t_\ell, t_r$ and the receiver side-information parameters $r_\ell, r_r$ in the sense that increasing/decreasing $t_\ell$ or $t_r$ by a positive integer $δ$ has the same effect on the multiplexing gain as increasing/decreasing $r_\ell$ or $r_r$ by $δ$. Moreover---even in asymmetric networks---there is an equivalence between the left side-information parameters $t_\ell, r_\ell$ and the right side-information parameters $t_r, r_r$.
We study the capacity of the discrete-time Gaussian channel when its output is quantized with a one-bit quantizer. We focus on the low signal-to-noise ratio (SNR) regime, where communication at very low spectral efficiencies takes place. In this regime a symmetric threshold quantizer is known to reduce channel capacity by a factor of 2/pi, i.e., to cause an asymptotic power loss of approximately two decibels. Here it is shown that this power loss can be avoided by using asymmetric threshold quantizers and asymmetric signaling constellations. To avoid this power loss, flash-signaling input distributions are essential. Consequently, one-bit output quantization of the Gaussian channel reduces spectral efficiency. Threshold quantizers are not only asymptotically optimal: at every fixed SNR a threshold quantizer maximizes capacity among all one-bit output quantizers. The picture changes on the Rayleigh-fading channel. In the noncoherent case a one-bit output quantizer causes an unavoidable low-SNR asymptotic power loss. In the coherent case, however, this power loss is avoidable provided that we allow the quantizer to depend on the fading level.
The source-coding problem with side information at the decoder is studied subject to a constraint that the encoder---to whom the side information is unavailable---be able to compute the decoder's reconstruction sequence to within some distortion. For discrete memoryless sources and finite single-letter distortion measures, an expression is given for the minimal description rate as a function of the joint law of the source and side information and of the allowed distortions at the encoder and at the decoder. The minimal description rate is also computed for a memoryless Gaussian source with squared-error distortion measures. A solution is also provided to a more general problem where there are more than two distortion constraints and each distortion function may be a function of three arguments: the source symbol, the encoder's reconstruction symbol, and the decoder's reconstruction symbol.
We derive the capacity region of the state-dependent semideterministic broadcast channel with noncausal state-information at the transmitter. One of the two outputs of this channel is a deterministic function of the channel input and the channel state, and the state is assumed to be known noncausally to the transmitter but not to the receivers. We show that appending the state to the deterministic output does not increase capacity. We also derive an outer bound on the capacity of general (not necessarily semideterministic) state-dependent broadcast channels.
We study channel capacity when a one-bit quantizer is employed at the output of the discrete-time average-power-limited Rayleigh-fading channel. We focus on the low signal-to-noise ratio regime, where communication at very low spectral efficiencies takes place, as in Spread Spectrum and Ultra-Wideband communications. We demonstrate that, in this regime, the best one-bit quantizer does not reduce the asymptotic capacity of the coherent channel, but it does reduce that of the noncoherent channel.
We propose a new inner bound on the capacity region of a memoryless multiple-access channel that is governed by a memoryless state that is known strictly causally to the encoders. The new inner bound contains the previous bounds, and we provide an example demonstrating that the inclusion can be strict. A variation on this example is then applied to the case where the channel is governed by two independent state sequences, where each transmitter knows one of the states strictly causally. The example proves that, as conjectured by Li et al., an inner bound that they derived for this scenario can indeed by strictly better than previous bounds.
We study the behavior of channel capacity when a one-bit quantizer is employed at the output of the discrete-time average-power-limited Gaussian channel. We focus on the low signal-to-noise ratio regime, where communication at very low spectral efficiencies takes place, as in Spread-Spectrum and Ultra-Wideband communications. It is well know that, in this regime, a symmetric one-bit quantizer reduces capacity by 2/pi, which translates to a power loss of approximately two decibels. Here we show that if an asymmetric one-bit quantizer is employed, and if asymmetric signal constellations are used, then these two decibels can be recovered in full.
An encoder observes a point pattern---a finite number of points in the interval $[0,T]$---which is to be described to a reconstructor using bits. Based on these bits, the reconstructor wishes to select a subset of $[0,T]$ that contains all the points in the pattern. It is shown that, if the point pattern is produced by a homogeneous Poisson process of intensity $λ$, and if the reconstructor is restricted to select a subset of average Lebesgue measure not exceeding $DT$, then, as $T$ tends to infinity, the minimum number of bits per second needed by the encoder is $-λ\log D$. It is also shown that, as $T$ tends to infinity, any point pattern on $[0,T]$ containing no more than $λT$ points can be successfully described using $-λ\log D$ bits per second in this sense. Finally, a Wyner-Ziv version of this problem is considered where some of the points in the pattern are known to the reconstructor.
We consider the transmission of a single bit over the continuous-time Poisson channel with noiseless feedback. We show that to send the bit reliably requires, on the average, half the energy of a photon. In the absence of peak-power constraints this holds irrespective of the intensity of the dark current. We also solve for the energy required to send $log_{2} M$ bits.
It is demonstrated that doubling the sampling rate recovers some of the loss in capacity incurred on the bandlimited Gaussian channel with a one-bit output quantizer.
The two-user discrete memoryless state-dependent multiple-access channel (MAC) models a scenario in which two encoders transmit independent messages to a single receiver via a MAC whose channel law is governed by the pair of encoders' inputs and by an i.i.d. state random variable. In the cooperative state-dependent MAC model it is further assumed that Message 1 is shared by both encoders whereas Message 2 is known only to Encoder 2 - the cognitive transmitter. The capacity of the cooperative state-dependent MAC where the realization of the state sequence is known non-causally to the cognitive encoder has been derived by Somekh-Baruch et. al. In this work we dispense of the assumption that Message 1 is shared a-priori by both encoders. Instead, we study the case in which Encoder 2 cribs causally from Encoder 1. We determine the capacity region for both, the case where Encoder 2 cribs strictly causal and the case where Encoder 2 cribs causally from Encoder 1.
New achievable rate regions are derived for the two-user additive white Gaussian multiple-access channel with noisy feedback. The regions exhibit the following two properties. Irrespective of the (finite) Gaussian feedback-noise variances, the regions include rate points that lie outside the no-feedback capacity region, and when the feedback-noise variances tend to 0 the regions converge to the perfect-feedback capacity region. The new achievable regions also apply to the partial-feedback setting where one of the transmitters has a noisy feedback link and the other transmitter has no feedback at all. Again, irrespective of the (finite) noise variance on the feedback link, the regions include rate points that lie outside the no-feedback capacity region. Moreover, in the case of perfect partial feedback, i.e., where the only feedback link is noise-free, for certain channel parameters the new regions include rate points that lie outside the Cover-Leung region. This answers in the negative the question posed by van der Meulen as to whether the Cover-Leung region equals the capacity region of the Gaussian multiple-access channel with perfect partial feedback. Finally, we propose new achievable regions also for a setting where the receiver is cognizant of the realizations of the noise sequences on the feedback links.
The continuous-time, peak-limited, infinite-bandwidth Poisson channel with spurious counts is considered. It is shown that if the times at which the spurious counts occur are known noncausally to the transmitter but not to the receiver, then the capacity is equal to that of the Poisson channel with no spurious counts. Knowing the times at which the spurious counts occur only causally at the transmitter does not increase capacity.
We study the best exponential decay in the blocklength of the probability of error that can be achieved in the transmission of a single bit over the Gaussian channel with an active noisy Gaussian feedback link. We impose an \emph{expected} block power constraint on the forward link and study both \emph{almost-sure} and \emph{expected} block power constraints on the feedback link. In both cases the best achievable error exponents are finite and grow approximately proportionally to the larger between the signal-to-noise ratios on the forward and feedback links. The error exponents under almost-sure block power constraints are typically strictly smaller than under expected constraints. Some of the results extend to communication at arbitrary rates below capacity and to general discrete memoryless channels.
We consider the transmission of a memoryless bivariate Gaussian source over an average-power-constrained one-to-two Gaussian broadcast channel. The transmitter observes the source and describes it to the two receivers by means of an average-power-constrained signal. Each receiver observes the transmitted signal corrupted by a different additive white Gaussian noise and wishes to estimate the source component intended for it. That is, Receiver~1 wishes to estimate the first source component and Receiver~2 wishes to estimate the second source component. Our interest is in the pairs of expected squared-error distortions that are simultaneously achievable at the two receivers. We prove that an uncoded transmission scheme that sends a linear combination of the source components achieves the optimal power-versus-distortion trade-off whenever the signal-to-noise ratio is below a certain threshold. The threshold is a function of the source correlation and the distortion at the receiver with the weaker noise.