On the Limits of Coexisting Coverage and Capacity in Multi-RAT Heterogeneous Networks

Chun-Hung Liu, Hong-Cheng Tsai
2017 IEEE Transactions on Wireless Communications  
This paper devises a general modeling and analyzing framework for a heterogeneous wireless network (HetNet) in which several wireless subnetworks coexist and use multiple radio access technologies (multi-RATs). The coexisting coverage and network capacity in such a multi-RAT HetNet are hardly investigated in prior works. To characterize the coexisting interactions in a multi-RAT HetNet, in this paper we consider a HetNet consisting of K tiers of APs and two different RATs, RAT-L and RAT-U, are
more » ... dopted in the HetNet. RAT-L is adopted by the access points (APs) in the first K − 1 tiers and APs in the Kth tier only use RAT-U. Both noncrossing-RAT and crossing-RAT user association scenarios are considered. In each scenario, the void probability and channel access probability of the APs in each tier are first found and then the tight lower bounds and their lowest limits on the proposed coexisting coverage and network capacity are derived. We show that multi-RAT networks in general can achieve higher link coverage and capacity by using opportunistic CSMA/CA that avoids/alleviates severe interfering between all coexisting APs. Also, crossing-RAT user association is shown to achieve much higher coexisting coverage and network capacity than noncrossing-RAT user association. Finally, numerical simulations for the LTE-U and WiFi networks coexisting in the HetNet validate our findings. A. Motivation and Prior Work A few earlier prior works on investigating the coexistence issue in multiple wireless networks mainly focused on how to efficiently and fairly share the unlicensed bands. In [7], a gametheoretical approach was proposed to solve the spectrum sharing problem for multiple coexisting and interfering networks. References [8]-[10] characterized the interference modeling and mitigation in the unlicensed bands. These works are not developed in a large-scale network model and the fundamental coexisting issues, such as the success transmission problem and network throughput, are not studied. A more accurate interference analysis technique based on the continuum field approximation and spiral representation was proposed in [11] for large-scale networks, but it still does not characterize the fundamental relationship between the interference and the intensities (density) of the wireless APs using different RATs. Recently, the coexistence problem in multi-RAT wireless networks has been gained more and more attentions since studying this problem helps different RAT networks jointly improve their wireless resource utilization. For example, LTE and WiFi networks can coexist in the unlicensed band to significantly improve the network capacity [12], [13] . To alleviate the coexisting interference impact in these two different kinds of wireless networks, the effective approach is either to offload traffic from LTE to WiFi networks or to make these two systems share the unlicensed spectrum resource in an appropriate way (see a recent work in [14] for this study). Purely offloading traffic from a LTE network to another WiFi network could not effectively improve the total capacity of these two networks when the WiFi network has limited resource for external offloading. On the contrary, if these two networks can coexist without causing severe interference, their sum capacity can be significantly improved. Reference [15] showed that small cell BSs have a notable throughput gain if they can adaptively access the unlicensed band without affecting the WiFi APs. In [16]-[19], stochastic geometry is applied to analyze the coexistence performance of large-scale LTE and WiFi networks, but the network models in these works are too simple to completely characterize the discrepancies originating from different RATs, such as distinct channel access protocols, different user association schemes for different RATs, etc. Hence, the analytical results in these works may be far away from their corresponding realistic outcomes.
doi:10.1109/twc.2017.2675399 fatcat:tkgo3jw3i5cddh3a7blphlavkm