Abstract
Emerging vertical applications enabled by connected devices and smart infrastructures have created an ever-increasing demand for high data rates over 5th-Generation (5G) and beyond wireless networks. Deployment of dense small cells (SCs) and millimeter wave (mmWave) communication systems have become inevitable in future wireless networks. Consequently, it is more accurate to model such networks in the 3D space due to the spatially distributed nature of the SCs, locations of the devices, radio resources and propagation environment. Accurate estimation of location-specific path loss parameters is then essential for efficient utilization of radio resources and management of dynamic coverage in 3D SC networks. In the paper, a framework for location-specific path loss estimation is developed for efficient radio resource management, based on the principle of crowdsensing together with Linear Algebra (LA) and machine learning (ML) techniques considering 2.5 GHz and 28 GHz bands. The corresponding procedure for capturing dynamic coverage of a SC base station (BS) serving to an arbitrary cluster is proposed and examined based on its 3D propagation characteristics. Results show that the accuracy of 3D channel parameter estimation using gradient descent ML techniques is superior compared to LA technique and can achieve over 98% estimation accuracy. It is shown that using the proposed process, parameters can be extrapolated for the slightly extended 3D communication distances from the cluster boundary for the worst-case locations of devices based on already estimated propagation parameters with accuracy over 74% for certain distances. Although numerical results are presented for a single amorphous 3D cell of a wireless network, the framework given in the paper can be extended to any arbitrary 3D wireless cellular network.
Highlights
T HE data traffic over wireless networks has been increasing dramatically over the past several decades and statistics show that globally this traffic increases by nearly a 1,000 times every 15 years [1]
For the implementation both 4G LongTerm Evolution (LTE) system operated in 2.5 GHz band [19], [20] and 5G new radio (NR) system and 28 GHz band [5], [40] are considered where the operating frequency 2.5 GHz is very closely related to the bands 7 and 53 [19] that are used in certain variants of LTE networks in different geographical regions
By recognizing necessity of utilization of third special dimension for cellular network planning in handling cell densification, wireless connections originated from different spatial poisons in 3D space and dynamic nature of the environment, location-specific path loss estimation and coverage management for dynamic 3D small cells (SCs) were done in this work
Summary
T HE data traffic over wireless networks has been increasing dramatically over the past several decades and statistics show that globally this traffic increases by nearly a 1,000 times every 15 years [1]. This work is primarily motivated by the technical problems associated with discovery of spatially distributed locationspecific path loss parameters in 3D cells where there is no universal set of propagation parameters, which can be used to estimate radio information at different locations or areas within a SC This location specific information is with a paramount importance for efficient utilization of radio resources to improve communication performance and for effective coverage management of devices for good QoS [23] while serving a large number of communication links. Sriyananda et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS extending the 3D communication distance at the border of a 3D cell or a cluster with extended RMs is considered as a critical aspect It is a process linked with the 3D spatial positions due to use of location specific propagation parameters while leading to an expansion of 3D SC coverage as a whole. The argument of the maximum value is given by arg max (·)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
