1.1.2. Capacity of Multiaccess Channels

Much of my work on the capacity of multiaccess channels has been focused on bridging the large gap between multiuser information theory and the channel models arising in modern multiuser communication systems such as CDMA (Code Division Multiple Access).

My first work on this area solved the open problem of the capacity of the multiple-access channel with memory (which plays a fundamental role in the analysis of the completely asynchronous channel):

S. Verdú, "Multiple-Access Channels with Memory with and without Frame-Synchronism,"
IEEE Trans. Information Theory, vol. IT-35, no. 3, p. 605-619, May 1989.

The issue of asynchronism is an important one in practical CDMA communication channels. Until the appearance of

S. Verdú, "The Capacity Region of the Symbol-Asynchronous Gaussian Multiple-Access Channel,"
IEEE Trans. Information Theory, vol. IT-35, no. 4, p. 733-751, July 1989

asynchronism had been studied in information theory at the codeword level, but not at the symbol level. That paper gives the capacity region of the additive white Gaussian noise power-constrained CDMA channel as a function of the signature waveforms assigned to the users. The capacity formula found in that paper and in

S. Verdú, "Capacity Region of Gaussian CDMA Channels: The Symbol-Synchronous Case,"
Proc. Twenty-fourth Allerton Conf. on Communication, Control and Computing, Allerton, IL, p. 1025-1034, Oct. 1986.

has been used by various authors to optimize the choice of signature waveforms. Optimal choices under RMS bandwidth constraints were obtained in

R. Cheng, S. Verdú, "Capacity of RMS Bandlimited Gaussian Channels,"
IEEE Trans. Information Theory, vol. IT-37, p. 453-465, May 1991.

Complete asynchronism reduces capacity by at most a factor equal to the number of users (a bound attained by TDMA), as shown in

R. Cheng, S. Verdú, "The Effect of Asynchronism on the Capacity of Gaussian Multiple-Access Channels,"
IEEE Trans. on Information Theory, vol. IT-38, p. 2-13, Jan. 1992

which also studies asynchronous multiplexing strategies that maximize capacity under bandwidth constraints.

Another application of the May 1989 paper on the capacity of multiaccess channels with memory was the solution of the capacity of the Gaussian multiaccess channel subject to intersymbol interference. This results in a nontrivial generalization of the classical single-user water-filling formula:

R. S. Cheng, S. Verdú,
"Gaussian Multiple-Access Channels with Intersymbol Interference: Capacity Region and Multiuser Water-Filling,"
IEEE Trans. on Information Theory, vol. IT-39, pp. 773-785, May 1993.

In 1971 a limiting expression for the capacity region of the memoryless interference channel was found. Since then, finding a single-letter characterization of that capacity region has remained one of the most famous open problems in multiuser information theory. Not even the Gaussian two-user interference channel has been solved. The formidable challenge in evaluating the limiting expression for such a channel is made evident by the main result in

R. S. Cheng, S. Verdú, "On Limiting Characterizations of Memoryless Multiuser Capacity Regions,"
IEEE Trans. on Information Theory, vol. IT-39, pp. 609-612, Mar. 1993.

which shows that restricting attention to input Gaussian processes incurs in loss of optimality even if they are allowed to have memory.

This page maintained by Michelle Young- Last modified 12/15/96