Dynamic Model for Network Selection in Next Generation HetNets With Memory-Affecting Rational Users

Abstract

Recently, due to the staggering growth of wireless data traffic, heterogeneous networks have drawn tremendous attention due to the capabilities of enhancing the capacity/coverage and to save energy consumption for the next generation wireless networks. In this paper, we study a long-run user-centric network selection problem in the 5G heterogeneous network, where the network selection strategies of the users can be investigated dynamically. Unlike the conventional studies on the long-run model, we incorporate the memory effect and consider the fact that the decision-making of the users is affected by their memory, i.e., their past service experience. Namely, the users select the network based on not only their instantaneous achievable service experience but also their past service experience within their memory. Specifically, we model and study the interaction among the users in the framework of fractional evolutionary game based on the classical evolutionary game theory and the concept of the power-law memory. We analytically prove that the equilibrium of the fractional evolutionary game exists, is unique and uniformly stable. We also numerically demonstrate the stability of the fractional evolutionary equilibrium. Extensive numerical results have been conducted to evaluate the performance of the fractional evolutionary game. The numerical results have revealed some insightful findings. For example, the user in the fractional evolutionary game with positive memory effect can achieve a higher cumulative utility compared with the user in the fractional evolutionary game with negative memory effect. Moreover, the fractional evolutionary game with positive memory effect can reduce the loss in the user's cumulative utility caused by the small-scale fading. 2 Network selection, fractional evolutionary game, memory-affecting rationality, and heterogeneous network.

User-Centric Radio Access Technology Selection: A Survey of Game Theory Models and Multi-Agent Learning Algorithms

“…Both issues are recently addressed in [149], that proposes the use of fractional eGT in a RAT selection scenario involving macrocells and mmWave small cells, some of it being provided by unmanned aerial vehicles (UAVs). Differently from classic eGT, fractional eGT includes memory of players in learning dynamics.…”

“…A stochastic game is formally represented by the tuple G S := {N , S, ∆(S), {A n } n∈N , {u n } n∈N }, where S represents the state space with elements s ∈ S. ∆(S) is a probability distribution over S, containing the probabilities p(s |s, a) of switching from state s to state s given that a strategy profile a is applied by players. ∆(S) is Markovian, since the next state s only depends on a and the previous 14 See [149] for more details. state s. 15 ∆(S) is also stationary, since it does not change over time.…”

The Meta Distributions of the SIR/SNR and Data Rate in Coexisting Sub-6GHz and Millimeter-Wave Cellular Networks

“…(26) in step (e) of Appendix C, we then obtain the expression for M −1,T as in Eq. (27) at the top of the next page.…”

“…Nakagami-m fading is a generic distribution that includes Rayleigh distribution (for non-LOS fading) as its special case when m = 1 and can well approximate the Rician fading distribution for 1 ≤ m ≤ ∞ (for LOS fading).3 Given highly directional beams and high sensitivity to blockage, recent studies showed that mm-wave networks can be considered as noise limited rather than interference limited[10],[23]-[27].…”