Precoder design for weighted sum delay minimization in MIMO physical layer multicasting

Hao Zhu, Narayan Prasad, Sampath Rangarajan

Research output: Chapter in Book/Report/Conference proceedingConference contribution


This paper considers the design of linear transmit precoding schemes to minimize the weighted sum delay metric over a K-user multi-antenna multicast channel. Limited by the rank and power constraints, the precoding matrices are designed under two interesting scenarios. The first scenario assumes the availability of pilots that can be precoded, using which the transmitter can convey any choice of transmit precoders to the users. Consequently, the sought transmit precoders can be any complex-valued matrices subject to given rank (dimensionality) and power (norm) constraints. A provably convergent cyclic alternating ascent based algorithm is proposed for a relaxed version of the problem, and is shown to attain at least a stationary point. Assuming that no such pilots are available, the second scenario constrains the transmit precoders to lie in a finite codebook. A concatenation based approach is adopted for constructing higher rank precoding matrices, which can facilitate the precoder search and allow for efficient signaling. A simple deterministic algorithm is proposed which involves maximizing a submodular rate function per step, and yields a worst-case performance guarantee.

Original languageEnglish (US)
Title of host publication2012 IEEE Global Communications Conference, GLOBECOM 2012
Number of pages6
StatePublished - 2012
Event2012 IEEE Global Communications Conference, GLOBECOM 2012 - Anaheim, CA, United States
Duration: Dec 3 2012Dec 7 2012

Publication series

NameGLOBECOM - IEEE Global Telecommunications Conference


Other2012 IEEE Global Communications Conference, GLOBECOM 2012
Country/TerritoryUnited States
CityAnaheim, CA

ASJC Scopus subject areas

  • Electrical and Electronic Engineering


Dive into the research topics of 'Precoder design for weighted sum delay minimization in MIMO physical layer multicasting'. Together they form a unique fingerprint.

Cite this