User Association for Millimeter-wave Enabled Cellular Networks

User Association for Millimeter-wave Enabled Cellular Networks

Project Overview:

User association is the problem of connecting users to base stations in an optimal way to increase the network performance. Max-SINR user association is a traditional user association method which uses the highest signal to interference plus noise ratio (SINR) to associate a user with a base station. This technique has been working well in cellular networks at microwave frequencies where all cells are homogeneous macro cells. The emergence of HetNets with smaller, lower power pico, femto base stations requires a different look at user association to ensure load balancing across the network. As a result, user association should be done based on base stations’ capabilities. Load balancing user association is a promising scheme which optimally connects users to base stations based on their load (balance the base station loads) and maximize the network throughput.

Load Balancing User Association:

User association problems usually models as a mixed integer nonlinear programming (MINLP), which is known to be NP-hard due to its non-convex nonlinear structure and presence of integer variables. Suboptimal and often heuristic methods exist for solving such an MINLP. In a mmWave system channels vary fast and significantly in a very short time. Because of the need for a fast and efficient algorithm tailored for user association in these systems, we designed an efficient algorithm, called Worst Connection Swapping (WCS), to solve the user association problem and find a near-optimal solution for it.

Results:

Figure 1: Max SINR versus Load Balancing User Association

Figure 1 compares user association using max-SINR scheme and WCS algorithm. It can be seen from the figure that the proposed method perfectly balances the BSs’ load by pushing the overloading UEs from the congested BSs into the lightly loaded BSs.

Figure 2: Effect of CSI and Interference on User Association

Figure 2 shows that the proposed interference model improves the network throughput by 60%. Moreover, the new idea of considering instantaneous mmWave channel variations revealed an improvement of almost 84% in network throughput. Also, it is shown that the proposed WCS algorithm significantly outperforms the traditional user association approaches, and it can increase the network throughput by optimal association of users to base stations.

Publications:

  1. “Load Balancing User Association in Millimeter Wave MIMO Networks,”
    A. Alizadeh and Mai Vu, IEEE Transactions on Wireless Communications, March 2019.
  2. “Time-Fractional User Association in Millimeter Wave MIMO Networks,”
    A. Alizadeh and Mai Vu, IEEE International Conference on Communications (IEEE ICC 2018), May 2018.