Uplink Coordinated Multi-Point in Field Trials and Ray Tracing Simulations Michael Grieger, Martin Danneberg, Gerhard Fettweis Vodafone Chair Mobile Communications Systems 01069 Dresden, Germany {michael.grieger, martin.danneberg, fettweis}@tu-dresden.de Mohammad Amro, Adnan Landolsi, Salam A. Zummo Electrical Engineering Department, KFUPM Dhahran, Saudi Arabia mohammad.amro@ieee.org, andalusi@kfupm.edu.sa Jens Voigt Actix GmbH Dresden, Germany jens.voigt@actix.com Abstract—Coordinated Multi-Point (CoMP) detection and transmission promises great improvements in cellular network coverage and throughput. To make full advantage of CoMP, however, appropriate radio planning is required that considers decoding of UEs at clusters of multiple BSs. Therefore, planning tools need accurate models that capture multi-cell propagation characteristics. This paper presents the analysis of uplink joint detection CoMP using ray tracing (RT). The main contribution is a comparison of RT simulation results with those of transmission experiments obtained through field trials. For this purpose, the testbed was modeled in a RT simulator. Measured and simulated results were evaluated for conventional and joint detection. I. I NTRODUCTION Cellular system design and standardization relies on ex- tensive simulation studies in modeled environments. This approach is powerful when the network is developed in in- cremental steps. Simulation models and their parameterization can be established and validated in existing real system imple- mentations. The already built-up experience is a trust base for innovations. The approach, however, is problematic when sig- nificant aspects of the cellular system structure are envisioned to be re-engineered in a rather revolutionary manner. In this case, experience shows that available models lack important aspects of real world signal propagation and underestimate technology challenges and constraints [1]. CoMP techniques completely change the fundamental notion that each user equipment (UE) communicates with one particular serving base station (SBS) [2]. Instead, signals of multiple UEs are jointly detected in a cluster of multiple coordinated base stations (BSs). One reliable method for proofing the practi- cability of an ambitious communication scheme and building empirical reference for its performance are field trials using prototype equipment for experimenting with true physical channels including the analog front-end. If successful, such field trials show that problems and challenges are solvable, which ignites further innovation in the industry. One drawback of field trials (FTs) is the extensive effort required to build the systems and perform measurement cam- paigns. Another is the usage of a particular hardware for UEs and BSs. Therefore, this approach is limited to the testing in specific environments and setups. A FT published in [3] showed that CoMP improves spectral efficiency by about 50 % and cell-edge user throughput by about 100 % compared to non-CoMP Conventional (Conv.) MIMO systems. As a major Fig. 1: FT setup and measurement trajectory, indicating the SBS ID at each location, and a detail of RT simulation in Actix RPS. Map data c Sandstein Neue Medien GmbH (http://stadtplan.dresden.de) limitation of the employed FT setup, the UEs were collocated on a measurement van in about 5 m distance. An alternative for obtaining insight into signal propagation is ray tracing (RT) [4]. RT captures multipath components in a 3D modeled replica of the environment [5], without any constraints concerning the antennas or the equipment. There- fore, RT models have the advantage of providing deployment- specific information without the need of on-site measurements or labor-intensive drive-test based model tuning [6]. RT was used in the design and simulation of different MIMO-based systems such as LTE and WiMAX [5]. As a downside, the accuracy of the environmental model and the RT propagation simulation is limited. This paper compares RT simulation results with real measurements. For this purpose, the real world testbed was modeled in a ray tracing simulator. Unlike [3], in the field trials two UEs were not collocated in close distance but moved freely in a mutual distance of about 5- 200 m. II. FIELD TRIAL SETUP The FT setup which consists of M = 13 BSs deployed at 5 sites in downtown Dresden is shown in Figure 1. It is located in downtown Dresden a representative area of a medium-sized