Radio Resource Sharing in Hierarchical Cellular Systems

Abstract: Hierarchical Cell Structures (HCSs) is generally believed tobe a solution to capacity problems of future mobilecommunication systems. In such systems we have a mix of cellsizes that handle different types of traffic. We can have largecells providing wide area coverage and small cells providinghigh capacity. Large macro cells serve fast moving users tominimize the number of handoffs (changing base station) in thesystem, small micro cells provide very high capacities per areaunit. The key question in these scenarios is how to share thescarce frequency spectrum between layers. Splitting thespectrum between layers results in lower trunking efficiency(lower number of users per frequency resource) but sharing thesame spectrum in several layers results in severe inter-layerinterference.We investigate different strategies for sharing the radiospectrum between layers in an HCS. These strategies are spreadspectrum sharing, i.e., DS-CDMA among macro users and TDMAamong micro users (System I) and vice versa (System II),DS-CDMA in both layers (System III). The commonly usedfrequency splitting (i.e., dedicated bands for each layer,System IV). We also investigate reuse of macro frequencies inunderlying systems (System V) and Slow Frequency Hopping inboth layers (System VI). For frequency splitting we make ananalysis of the effects of Adjacent Channel Interference on thecapacity. We compare all systems in terms of the capacity,i.e., the supported number of users per cell. We develop a newanalytical method for the capacity analysis of mobilecommunication systems, including the effects of soft handoffs,log-normal fading, thermal noise and constant received powercontrol.The analysis show that quality based power control canhandle the varying layer external interference in HCSs. Becauseof the severe inter-layer interference is the capacity tradeoff bad for all systems sharing the same spectrum in bothlayers (System I, II, III and VI) except when reusing macrofrequency channels in underlying layers (System V). System Vperforms better than the frequency splitting (System IV) in theup-link, for the down-link is the performance of these twoschemes close. The results indicates that the severeinter-layer interference can not be balanced by the trunkinggain from not splitting the spectrum.