Abstract
Spectrum allocation within the fixed unlicensed band affects performance of wireless networks. Fundamental limitations of spectral efficiency on capacity of wireless local area networks (WLANs) hence needs to be studied. Recently, it was shown that the performance can be increased when both non-overlapping and partially overlapping channels are used. Unlike previous studies, this is the first known attempt on analysis of information theoretic capacity for partially-overlapping channels, as characterized by IEEE 802.11b-type systems using a hyper-receiver. Using Marenko-Pastur law distribution, capacity is approximated for such systems. We implement this by proposing a novel channel model for our analysis. The results conform to Monte Carlo simulations, with an approximate mean error of less than 5%. Fundamental tradeoffs of channel interference and power gain are discussed. We further explore the effect of access point density and find that systems based on partially overlapping channels perform well in high density deployments. We also compute capacity for a range of channel overlaps and conclude that rates close to capacity are achievable for a channel overlap of 90% and above.