|| On Optimal Power Allocation for Modulation-Constrained Gaussian Channels
||Yu Han, Maria Urlea, Sergey Loyka, University of Ottawa, Canada|
||D7-S3-T2: Optical & Wireless Communications
||Tuesday, 20 July, 22:40 - 23:00
||Tuesday, 20 July, 23:00 - 23:20
The problem of optimal power allocation for parallel Gaussian channels under modulation order constrains, in addition to the total transmit power constraint, is considered. It is motivated by coded-modulation systems using powerful capacity-approaching codes. While only analytically-intractable solution is known to this problem, an explicit closed-form solution is obtained here using a sphere-packing-based approximation for modulation-constrained rates. It can be interpreted as waterfilling with variable water level, which is also expressed in a closed-form. The obtained power allocation also solves the dual problem of minimizing the total transmit power subject to the sum rate and modulation order constraints. More insightful analytical solutions are obtained in some special cases. While the new power allocation is similar to the well-known waterfilling procedure at low SNR, it is dramatically different at moderate to high SNR. Proportional cardinality allocation is shown to be optimal at high SNR under the uniform power allocation.