TY - GEN
T1 - Iterative Reweighted Least Squares approach to interference alignment
AU - Rihan, Mohamed
AU - Elsabrouty, Maha
AU - Elnouby, Said
AU - Shalaby, Hossam
AU - Muta, Osamu
AU - Furukawa, Hiroshi
PY - 2013
Y1 - 2013
N2 - This paper investigates the interference alignment (IA) solution for a K-user static flat-fading multiple input multiple output (MIMO) interference channel. Optimal users' precoders and postcoders are designed through a rank constraint rank minimization (RCRM) framework with IA conditions inserted within the constraints and the cost function of a complex matrix optimization problem. With RCRM formulation, the interference is forced to span the lowest dimensional subspace possible, under the condition that the useful signal subspaces span all available spatial dimensions. Using the recent advances in matrix completion theory and low rank matrix recovery theory, we propose an Iterative Reweighted Least Squares (IRLS) approach to IA. Through this approach, we provide an adequate relaxation for the rank function which in some cases attain the same results obtained using the standard nuclear norm with lower elapsed time per iteration and lower number of iterations and in some cases perform better than any of the previous approaches.
AB - This paper investigates the interference alignment (IA) solution for a K-user static flat-fading multiple input multiple output (MIMO) interference channel. Optimal users' precoders and postcoders are designed through a rank constraint rank minimization (RCRM) framework with IA conditions inserted within the constraints and the cost function of a complex matrix optimization problem. With RCRM formulation, the interference is forced to span the lowest dimensional subspace possible, under the condition that the useful signal subspaces span all available spatial dimensions. Using the recent advances in matrix completion theory and low rank matrix recovery theory, we propose an Iterative Reweighted Least Squares (IRLS) approach to IA. Through this approach, we provide an adequate relaxation for the rank function which in some cases attain the same results obtained using the standard nuclear norm with lower elapsed time per iteration and lower number of iterations and in some cases perform better than any of the previous approaches.
UR - http://www.scopus.com/inward/record.url?scp=84893633870&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84893633870&partnerID=8YFLogxK
U2 - 10.1109/WD.2013.6686521
DO - 10.1109/WD.2013.6686521
M3 - Conference contribution
AN - SCOPUS:84893633870
SN - 9781479905423
T3 - IFIP Wireless Days
BT - 2013 IFIP Wireless Days, WD 2013
PB - IEEE Computer Society
T2 - 6th IFIP/IEEE Wireless Days Conference, WD 2013
Y2 - 13 November 2013 through 15 November 2013
ER -