1: \begin{abstract}

2: 

3: \noindent This paper studies the  \emph{instantaneous rate} maximization and the \emph{weighted sum delay} minimization problems over a $K$-user multicast channel, where multiple antennas are available at the  transmitter as well as at all the receivers. Motivated by  the degree of freedom optimality and the simplicity offered by linear precoding schemes, we consider the design of linear precoders using the aforementioned two criteria. We first consider the scenario wherein the linear precoder can be any complex-valued matrix subject to rank and power constraints. We propose   cyclic alternating ascent based precoder design algorithms and establish their convergence to respective stationary points. Simulation results reveal that our proposed algorithms considerably outperform known competing solutions.

4: We then consider a scenario in which the linear precoder can be formed by selecting and concatenating precoders from a given finite codebook of  precoding matrices, subject to  rank and power constraints. We show that under this scenario, the instantaneous rate  maximization problem  is equivalent to a robust submodular maximization problem which is strongly NP hard. We propose a deterministic approximation algorithm and show that it yields a bicriteria approximation. For the weighted sum delay minimization problem  we propose a simple deterministic greedy algorithm, which at each step entails approximately maximizing a submodular set function subject to multiple knapsack constraints, and establish its performance guarantee. %then consider a relaxed version of the weighted sum delay minimization problem, which while NP-hard enables constant-factor approximation algorithms.

5: \end{abstract}

6: