Spectral Efficiency Analysis with Channel-aware Schedulers
Accurate planning and performance evaluation of current and next generation cellular wireless communication systems requires understanding the joint impact of scheduling, interference, and fading. This project strives to provide accurate analytical expressions for the spectral efficiency in bits/sec/Hz of cellular systems that use either channel-aware of channel-unaware schedulers.
Background & Objective: While considerable amount of work has been done in the literature on the spectral efficiency of cellular systems, it either requires costly numerical simulations or makes several simplifying assumptions that limit its applicability. We develop a general analysis that allows for non-identical co-channel interference from neighboring cells, and in which all links undergo both small-scale Rayleigh fading and large-scale lognormal shadowing. Doing so avoids the loose bounding approaches of the literature that assume that interferers are located in the worst-case or best-case locations. The effect of finite modulation constellations on spectral efficiency is also modeled. Both the fair, but channel-unaware, round-robin scheduler and the greedy, but unfair, max-throughput or max-SIR scheduler are analyzed.
Technical Discussion: Our analysis incorporates two additional tweaks, which are essential in making it more accurate than previous works: (i) the use of a novel moment generating function based lognormal approximation method instead of the conventional Fenton-Wilkinson approximation approach, and (ii) an accurate approximation of the Gaussian-Q function instead of a looser bound. Consequently, our analytical results match the reference simulation results very well. Our results demonstrate that multi-tier interference has a greater impact on the performance of the round-robin scheduler than the max-throughput scheduler. On the other hand, the limited constellation size affects the max-throughput scheduler much more than the round-robin scheduler, especially when the users are close to the serving base station. Our analysis thus provides a useful benchmark for calibrating system-level simulators, which are required to evaluate the overall multi-user, multi-cell performance of cellular systems.
Outside Collaborations: Prof. Jingxian Wu, Sonoma State University, Sonoma, CA.
Future Direction: Future work includes generalizing the analysis to proportional fair schedulers, which effectively balance throughput and fairness.
Contacts:
Andreas F. Molisch
Jinyun Zhang
| Technical Reports: | |
| Spectral Efficiency Analysis of Cellular Systems with Channel-Aware Schedulers | |
Technology Area: Digital Communications
Modification Date: September 12, 2007
