|Pranav Madadi||University of Texas at Austin, USA|
|Francois Baccelli||UT Austin & The University of Texas at Austin, USA|
|Gustavo De Veciana||The University of Texas at Austin, USA|
Ultra densification along with the use of wider bands at higher frequencies are likely to be key elements towards meeting the throughput/coverage objectives of 5G wireless networks. In addition to increased parallelism, densification leads to improved, but eventually bounded, benefits from proximity of users to base stations, while resulting in increased aggregate interference. Such networks are expected to be interference limited, and in higher frequency regimes, the interference is expected to become spatially variable due to the increased sensitivity of propagation to obstructions and the proximity of active interferers. This paper studies the characteristics of the spatial random fields associated with interference and Shannon capacity in ultra-dense limiting regimes. They rely on the theory of Gaussian random fields which arise as natural limits under densification. Our models show how densification and operation at higher frequencies, could lead to increasingly rough temporal variations in the interference process. This is characterized by the Hölder exponent of the interference field. We show that these fluctuations make it more difficult for mobile users to adapt modulation and coding. We further study how the spatial correlations in users' rates impact backhaul dimensioning. Therefore, this paper identifies and quantifies challenges associated with densification in terms of the resulting unpredictability and the correlation of interference on the achievable rates.