說明 
118 p 
附註 
Source: Dissertation Abstracts International, Volume: 6709, Section: B, page: 5291 

Adviser: John M. Cioffi 

Thesis (Ph.D.)Stanford University, 2006 

The broadcast channel is an informationtheoretic model of a communication system whereby a transmitter communicates to a number of receivers. These receivers, because of their geographical locations or other practical limitations cannot cooperate in decoding their individual messages. Examples of such a channel model are downlink communication in a wireless cellular network or downstream transmission in a Digital Subscriber Line system. Certainly, a full understanding of the communication limits over broadcast channels provides practical insights on the design of more efficient wireless or wireline networks 

In this dissertation, an alternative converse proof for the capacity region of the Gaussian vector broadcast channel is presented. First, the achievable region by the dirty paper coding (DPC) scheme is reviewed, and the multipleaccess broadcast channel duality is revisited. By combining some ideas from earlier works on the problem with duality, primarily, the optimality of the DPC scheme is proven under a total transmitpower constraint. Afterward, by employing a more comprehensive notion of the multipleaccess broadcast channel duality, the optimality proof for the DPC scheme is extended to general convex constraints on the transmit covariance matrix 

The capacity expression for the Gaussian vector broadcast channel is used subsequently to establish the capacity region for a family of parallel Gaussian vector broadcast channels with total average transmitpower constraint. It is shown that using independent codebooks over the individual subchannels is capacity achieving. Moreover, the problem of optimal power allocation over the subchannels is studied from two major informationtheoretic standpoints. In the first problem, the power allocation policy that uses the minimum total transmit power to achieve a given point in the capacity region is investigated. By transforming the underlying optimization problem to the Lagrange dual domain, an efficient numerical algorithm is proposed to find this power allocation policy. In the second problem, the policy that achieves a boundary point of the capacity region is studied. It is shown that this power allocation problem is essentially the same as the minimum power allocation problem in the dual domain. Based on this similarity, a numerical algorithm is developed to solve this power allocation problem 

School code: 0212 

DDC 
Host Item 
Dissertation Abstracts International 6709B

主題 
Engineering, Electronics and Electrical


0544

Alt Author 
Stanford University

