Cooperative Load Balancing on Human Moving Trajectory Diversity in Cellular Networks Kao-Peng Chou Department of Communication Engineering National Central University, Taiwan Email: 985403006@cc.ncu.edu.tw Jia-Chin Lin Department of Communication Engineering National Central University, Taiwan Email: jiachin@ce.ncu.edu.tw Abstract—Human dynamics are investigated for deployment of picocell in cellular network in this paper. Human dynamics may lead to variation of channel condition, e.g., moving fast or slow, driving in a mega city or in a countryside may change channel conditions. The data delivery throughput depends on user distribution, commuting trajectory, and Internet access regularity. Effectiveness of frequency reuse assisted from picocell deployment is affected by human dynamics. By applying human dynamics to a network management strategy, the effectiveness of frequency reuse can be improved in practice. Pricing strategy and resource management can be reconsidered according to users’ loyalty with respect to a service provider. Novel pricing strate- gies based on user’s request-to-receive ratio are also proposed. Computer simulations verify the improvements obtained using the proposed methods. Index Terms—Human dynamics, frequency reuse, human tra- jectory. I. I NTRODUCTION Mobile communication networks grow more and more complicated. Technologies, such as heterogeneous networks, Internet of Things, device-to-device communication and coor- dinated multipoint, have focused attention in recent years. A common goal of aforementioned technologies is to increase spectral efficiency. For example, in a heterogeneous network, various levels of basestations, such as macrocells, picocells and femtocells, coexist and cooperate in a single cell. A macrocell basestation is known as evolved node B (eNB) in the long-term evolution (LTE) specifications, which works as a usual basestation in previous generations of mobile com- munication. A picocell is deployed within macrocell coverage and can be accessed by any user, therefore, increasing the data transmission capacity in the cell. In a heterogeneous network, users can access the network in the same frequency if they subscribe either to different picocells or the macrocell. This also indicates that cell capacity may be related with human dynamics in practice. Human dynamics were involved with a branch of research in statistical physics, including crowd movement, human behaviors, and human interactions with respect to networks. Human move- ment trajectory research was often investigated by modeling human mobility in accordance with stochastic models, such as the Levy walk model [1] or Brownian motion. The limits of predictability in the human mobility obtained by observing cell phone calling records have been investigated [2]. The author surprisingly determined that travelers with different moving patterns have comparable predictability. Additionally, a previous study [3] categorize human characteristics into two categories in terms of mobility. Human movement trajectory and the data access request modeling are highly correlated with network efficiency and network dynamics. Network dynam- ics research systematically investigates complicated networks, which can be applied in many different fields, such as power grids, neural systems, and cellular network management. The complexity of a dynamic computer network is defined and evaluated by analyzing the distribution of an entropic degree [4], which is constrained by network satisfaction and network cost; the dynamics of a network are attributed to generate/delete nodes and links with certain probabilities. A previous study [5] investigated the dynamics of a mobile sensor network (MSN) in which nodes can move with a parallel moving algorithm to balance the net force. A previous study [6] explored the possibility of using designed admissible compensatory perturbations to stabilize the system. In this paper, users are considered as dynamic nodes in a network. Users can change the subscribed network according to human dynamics; the change influences the network dynamics. The objective is to control the network dynamics to optimize network efficiency. According to results provided by the previous study [2], a human movement trajectory is highly predictable. A strategy based on the trajectory controls a user to engage in network interaction. This may improve the network efficiency by taking advantage of trajectory diversity. To verify the dynamics, we propose a non-return dual-network structure in which two networks provide service via non- overlapping spectrum in the same cell. Users in the cell can change networks or leave the cell in a stochastic manner, i.e., based on probability distributions of service quality, service fee and tolerance of individuals. II. HUMAN DYNAMICS MODEL Methods based on spots-based map are commonly used in mobility management areas because of the sources of the movement information. Previous works [2], [7], [8] used cell phone log to identify user trajectory, which is considered as a series of connected spots regarding the nearest basestations. Define a network structure in a wireless cellular cell, in which users with the same service provider are regarded as nodes in IEEE ICC 2017 Wireless Communications Symposium 978-1-4673-8999-0/17/$31.00 ©2017 IEEE 6623