A New Deterministic Hybrid Model for Indoor-to-Outdoor Radio Coverage Prediction Dmitry Umansky *† , Guillaume de la Roche ‡ , Zhihua Lai § , Guillaume Villemaud *† , Jean-Marie Gorce *† , Jie Zhang ‡§ * CITI, INSA-LYON, F-69621, Villeurbanne, France † INRIA, Universit´ e de Lyon, F-69621, Villeurbanne, France ‡ CWiND, University of Bedfordshire, Park Square Campus, Luton, LU1 3JU, UK § Ranplan Wireless Network Design Ltd, 1 Kensworth Gate, Luton LU6 3KS, UK Abstract—In this article, we propose a new hybrid modeling method for indoor-to-outdoor radio coverage prediction. The pro- posed method is a combination of a ray-optical channel modeling approach and the frequency domain ParFlow method. While the former is widely used for modeling outdoor propagation environments, the latter is computationally efficient and accurate for modeling indoor environments. I. I NTRODUCTION The ubiquitous deployment of various wireless communi- cation networks, particularly in urban areas, requires careful planning of new wireless networks, as well as optimization of the existing ones. Successful accomplishment of these tasks calls for efficient radio network design tools. Unavoidably, any debate about merits and demerits pecu- liar to a concrete tool, or more precisely, to an underlying electromagnetic wave propagation modeling approach, leads to a discussion about the trade-off between the computational load and the achievable accuracy of the prediction. To a large extend, the compromise between efficiency and accuracy depends on the modeled propagation environment. It has been demonstrated that the multi-resolution frequency domain ParFlow (MR-FDPF) method [1] is an efficient and accurate radio network design tool for indoor and indoor-like environments. Yet the computational load associated with this method quickly becomes excessively large due to the size increase of the propagation environment as, for example, in outdoor wave propagation scenarios. On the other hand, the well-known ray-optical approaches [2], [3] are widely used for modeling the outdoor as well as indoor environments. Even so, using the ray-optical methods for accurate prediction of the electrical field strength inside a building might not be as computationally efficient as employing the MR-FDPF method (see also discussion in [4]). Moreover, MR-FDPF method is usually more accurate since it does not restrict the number of reflections to be computed as it is the case in ray-optical approaches. For scenarios where both the indoor and outdoor wave propagations have to be considered, a combination of the MR- FDPF and the ray-optical methods promises advantages in providing accurate prediction results without sacrificing the computational efficiency. Indeed, performing the simulation of the whole indoor-to-outdoor scenario based on MR-FDPF only would require too much memory. It is to be noticed that a combination of the ray-optical and the MR-FDPF methods for predicting the electrical field strength in outdoor-to-indoor wave propagation scenarios has been already explored in [5]. In this article, we propose a new method for combining the ray-optical and the MR-FDPF approaches in order to accurately and efficiently predict the field strength in indoor- to-outdoor wave propagation scenarios. Several works related to this subject can be found in [6] and the references therein. The rest of the paper is organized as follows. In Section II, we briefly introduce the principal problem associated with combining the ray-optical and the MR-FDPF approaches. The method that allows combining the two modeling approaches is described in Section III. The results of the performance eval- uation of the proposed hybrid modeling method are presented in Section IV. The concluding remarks are given in Section V. II. PROBLEM FORMULATION Before going into the details related to the development of the hybrid model, we briefly describe the different paradigms, which the MR-FDPF and the ray-optical modeling approaches are based on. In the ray-optical modeling method, the electrical field strength at each point is calculated as a sum of the rays passing through the point. Each ray obeys the laws of geometrical optics. The reflections and diffractions of transmitted signals on the obstacles are computed by tracing all the possible rays between transmitters and receivers. In contrast, the MR-FDPF method is based on the cellu- lar automata formalism [7]. The electrical field strength is obtained by summing the fictitious flows traveling along a regular grid of transmitting lines and experiencing scattering at the nodes of the grid. It follows that additional post-processing is required in order to transform the prediction results provided by the MR-FDPF method into a set of appropriate parameters of the rays to be used in the ray-optical methods. This will be described in the next section. III. COMBINATION OF MODELS As it has been mentioned above, the MR-FDPF method cannot be directly interfaced with the ray-optical methods. EuCAP 2011 - Convened Papers 3771