TY - GEN
T1 - Impact of interference coupling - loss of convexity
AU - Boche, Holger
AU - Siddharth, Naik
PY - 2009
Y1 - 2009
N2 - In interference coupled wireless systems, where it is not possible to "orthogonalize" all the users in the system, we characterize the impact of interference coupling on the convexity of certain utility functions and problems. We introduce a general class of competitive user utility functions and natural competitive user utility functions. We further introduce the signal-to-interference based utility functions, which are based on physical layer parameters in wireless systems. We prove the conditions, which when satisfied - result in a competitive user utility function being a signal-to-interference ratio based utility function. We further show that there exists no natural competitive user utility function, which is convex or concave. Furthermore, we show that a sum of weighted combination of natural competitive user utility functions is not convex or concave. Such functions are commonly encountered in wireless communication systems, e.g. rate or MMSE as a function of signal-to-interference ratio. We show that such rate maximization or MMSE minimization problems are not convex programs under our specified conditions.
AB - In interference coupled wireless systems, where it is not possible to "orthogonalize" all the users in the system, we characterize the impact of interference coupling on the convexity of certain utility functions and problems. We introduce a general class of competitive user utility functions and natural competitive user utility functions. We further introduce the signal-to-interference based utility functions, which are based on physical layer parameters in wireless systems. We prove the conditions, which when satisfied - result in a competitive user utility function being a signal-to-interference ratio based utility function. We further show that there exists no natural competitive user utility function, which is convex or concave. Furthermore, we show that a sum of weighted combination of natural competitive user utility functions is not convex or concave. Such functions are commonly encountered in wireless communication systems, e.g. rate or MMSE as a function of signal-to-interference ratio. We show that such rate maximization or MMSE minimization problems are not convex programs under our specified conditions.
UR - http://www.scopus.com/inward/record.url?scp=77951571399&partnerID=8YFLogxK
U2 - 10.1109/GLOCOM.2009.5426117
DO - 10.1109/GLOCOM.2009.5426117
M3 - Conference contribution
AN - SCOPUS:77951571399
SN - 9781424441488
T3 - GLOBECOM - IEEE Global Telecommunications Conference
BT - GLOBECOM 2009 - 2009 IEEE Global Telecommunications Conference
T2 - 2009 IEEE Global Telecommunications Conference, GLOBECOM 2009
Y2 - 30 November 2009 through 4 December 2009
ER -