A few important propagation scenarios based on the METIS (Mobile and wireless communications Enablers for the Twenty-twenty Information Society) test cases are briefly explained to address 5G specific modelling requirements.

A. Virtual Reality Office

In this TC huge amounts of data are exchanged to enable interactive work among people in remote locations (such as high-resolution 3D data that gives the amazing experience “as if you were there”). It is assumed that the users can rely on getting bitrates higher than 1 Gbps. The propagation challenge is to characterize higher frequencies (in the millimetre band), larger bandwidth and higher antenna directivity.

B. Dense urban information society

In this TC the challenge is to ensure connectivity at any place and at any time in dense urban areas for the traffic between humans and between human and the cloud. Here bitrates in the range 50-300 Mbps will be guaranteed for almost all users. The main propagation challenge here is to improve the 3D channel and path loss modelling. Moreover, full large scale mobility providing realistic correlations characteristic between different links is needed.

C. Traffic efficiency and safety

This TC addresses super real-time and reliable connections as illustrated in Fig. 1 where traffic accidents will be avoided by cooperative intelligent traffic systems that require timely andreliable exchange of information of less than 5 ms end-to-end (E2E) latency. The corresponding propagation challenge is to model the device to device (D2D) channel allowing for mobility in both ends of the link.

The METIS propagation modelling needed for developing the 5G mobile communications technology is largely based on existing and widely used geometry based stochastic channel models (GSCM) [4,5,8]. These models are popular partly because of their scalability and reasonable complexity. Some of the most important models are briefly described below.

A. WINNER/IMT-Advanced

These models describe a versatile set of environments, ranging from indoor to a variety of indoor-to-outdoor and outdoor. The parameterizations of the models are based on a fair number of channel measurement campaigns. The modelling framework supports three-dimensional description of environments. Early implementations have 2D parameterization only while contains an extension to elevation dimension. Channel realizations are generated by superimposing plane waves with certain characteristic parameters. Small scale parameters of rays are: departure and arriving angles, propagation delay and power. A continuous evolution of these parameters is not accounted for as the sub-paths (plane waves) are not based on geometric locations of scattering clusters. Instead, theparameters are chosen randomly from appropriate probability distributions. Large scale characteristics like rms delay spreadmay however evolve smoothly.

Quadriga is an extension of the WINNER II/+ models. It is a full 3D model with geometric polarization and it can be used for terrestrial as well as satellite communications. Continuous time evolution for links with a single base station is supported with environment transitions and small and large scale fading between segments.

The GSCMs are not spatially consistent meaning that they don’t fully support continuous motion beyond stationary intervals. The plane wave assumption (inherent in angular description) sets the following requirement: antenna arrays are assumed to be small enough such that all elements experience the same large scale parameters. This assumption may not be valid with 5G requirements. Another characteristic of GSCM is that they are mostly developed for cellular deployment with fixed base stations. Thus the models are not parameterized for peer-to-peer (D2D) links. Also the range of frequencies covered by the existing parameterizations is in the range of present cellular networks.

B. COST 2100

The COST 2100 model is closely related to the WINNER model as both have common origin. The COST 2100 model is however not restricted to the drop concept with short independent segments of motion. By modelling spatially located scattering clusters and their corresponding visibility regions the COST 2100 model supports non-stationary and continuous evolution of the radio channel. If a UE is within a visibility region of a cluster the signal propagates via that cluster. Two UEs close to each other are likely to see partly the same clusters, and thus are likely to have similar angular characteristics. There are two basic types of clusters: 1) single bounce clusters, and, 2) twin clusters that enable multi-bounce modelling with specific last and first bounce cluster co-ordinates. The COST model supports spherical wave modelling and also spatial consistency by means of geographically located clusters and their corresponding visibility regions. However, the COST does not account for mobility of both ends of the link and is thus is not applicable in D2D. Also parameterizing the COST model to different environments according to measurements is a challenge, because input parameters like cluster properties are not straightforward to extract from measurements.

C. IEEE 802.11 for 60 GHz

The IEEE 802.11ad channel model is intended for 60 GHz Wireless Local Area Networks (WLANs) where very high data rates are required. The model is cluster-based and describes the channel by providing accurate space-time characteristics including polarization and supports both beamforming and non-stationary characteristics of the channel. As a result of

experimental measurements and ray-tracing simulations, the model is parameterized for three indoor scenarios, namely a conference room, a cubicle and a living room. However, since the model parameters are determined deterministically, the parameterization for each scenario is site-specific and may not be valid for other similar environments.

There are two main factors determining requirements on the propagation modelling. The first is the scenarios from the environment and user perspective and the second is the technology components envisaged to provide the required end user services. The scenarios were briefly described in section II. From a technology perspective, the propagation challenge is Fig. 1. Traffic safety test case illustrating collision avoidance. Mobile D2Dmainly higher frequencies and wider bandwidths, together with much larger antenna arrays in terms of number of elements and in terms of physical size with respect to the wavelength. Combining these two factors the following main challenges have been identified.

A. Spatial consistency and mobility

The current most commonly used channel models are drop based, meaning that the scattering environment is randomly created for each link. The corresponding performance of spatial techniques like MU-MIMO is exaggerated, because even close-by mobiles see independent scatterers, which is not the case in reality. As the importance of spatial techniques, as well as the density of links is expected to increase, it is increasingly important to model these links in a consistent manner. A spatially consistent model can also inherently support mobility of users. To create a consistent model, geometric locations of the scatterers of the first and last hop of each path (transmitter-to-scatterer and scatterer-to-receiver) have to be defined. Moreover, a death and birth process of rays has to be defined according to the visibility of the scatterers. No known model can describe this dual mobility in a consistent manner.

B. Diffuse vs. specular scattering

Commonly used channel models assume scattering by geographically fixed clusters. This assumption is appropriate for diffuse scattering and diffraction. Literature [9] and measurements performed in METIS (Fig. 4) indicate, however, that specular paths may dominate in many scenarios. The characteristics of specular paths are very different from diffuse paths regarding apparent scatterer locations which are not fixed for specular propagation. As 5G transmission schemes are expected to utilize steerable highly directive and/or very large MIMO antennas the channel modelling should take into account realistic modelling of specular paths.

C. Very large antenna arrays

An important technology component of 5G mobile communications is the use of very large antenna arrays (which even may extend over large scale fading regions) for e.g. massive MIMO and pencil beamforming as illustrated in Fig. 2. For these highly directive antennas or large antenna arrays substantially non-realistic performance will be achieved using present modelling. Transmission schemes based on antenna arrays extending over many wavelengths (which does not imply a large physical array size at millimetre wavelengths) exploit super-resolved channel properties. Current channel modelling needs corresponding improvement in angular resolution as well as sub-path amplitude distribution. Further, these large arrays require spherical wave modelling replacing the commonly used plane wave approximation.

D. Millimetre wave frequencies

The millimetre wave frequencies have very promising prospects of providing substantial additional amount of both spectrum and spatial multiplexing capacity. Though millimetre propagation has been investigated quite extensively, particularly at 60 GHz, crucial characteristics such as highly resolved angular properties and NLOS path loss are not well known.

The final METIS channel model will be based on the previous European 6th framework project WINNER II, and Celtic project WINNER+ channel models to as large extent as possible. Required extensions and modifications are based on METIS measurement campaigns and literature. The main extensions/modifications concern frequency range, spatial consistency, 3D (elevation) and spherical waves. This work is on-going and an interim version of the METIS models will be available in May 2014, and the final model is expected to be ready by March 2015.

A. Frequency range

WINNER II/+ and IMT-Advanced models have been designed for frequency range below 6 GHz. The METIS model should cover frequencies from 380 MHz up to 86 GHz. Due to the extremely large frequency range and limited availability of channel sounders, only some “snapshots” from the range can be measured within METIS. The gaps between those snapshots are filled based on a literature review, software simulations, and interpolation. Our assumption is that the GSCM principle is applicable to the millimeter waves as well, but many parameter values depend on the frequency.

B. Spatial consistency

For obtaining the spatial consistency, it is necessary to define the geographical cluster location (x, y, z coordinates) and visibility region/cluster lifetime. For the single bounce case, the delay, Angle of Arrival and Departure (AoA, AoD) depend on each other according to the geometry (Fig. 3). However, in the case of multiple bounces, AoA, AoD, and delay are assumed to be independent. When the transmitter and/or receiver moves a short distance, the AoA, AoD and delay are adjusted based on the geometry. In a longer movement, cluster locations need to be updated depending on corresponding cluster visibility regions. When a device is moving, the clusters are updated individually according to their visibility regions. For the single bounce case the receiver and transmitter are within the same visibility

region. For multi-bounce each end of the link has its own visibility region. Fig. 2. Scenario using very large wall mounted antenna arrays.This approach enables both single mobility and dual mobility (moving BSs, D2D), smooth time evolution of small scale and large scale parameters. Additionally, it has been discussed whether clusters can be located according to the physical environment, e.g. Manhattan grid. This could be done either by defining each building as a cluster (or a group of clusters) or using almost rectangular type statistical distributions of angular parameters/cluster locations.

C. 3D Extension

The 3D extension of METIS model follows the principles described in WINNER+ [8] and 3GPP. Additionally over the rooftop diffraction in macro-cell cases could be modelled more accurately.

D. Spherical waves