Project Details

It is estimated that 66% of voice services and 90% of data services (e.g., Internet access) using wireless technologies take place indoors. How to plan and design good indoor radio networks has become a paramount issue. However, the vast majority of the work on Radio Access Network (RAN) Planning and Optimisation (P&O) has been done for outdoor rather than indoor. As a result, there is a lack of high quality indoor RAN P&O tools.

In this project, indoor radio network P&O methods and tools will be developed. The tools will provide some 20-30% performance improvements, reduce power usage and radio wave pollutions (thus reduce the health concerns) of indoor radio systems, and improve information security (due to the reduction of signal leakage).

The objectives of the project are as follows:

  • Research and develop fast and accurate radio propagation models that can be used for indoor radio network planning and optimisation.
  • Investigate various issues arising from the use of femtocells, for example, the study of interference between femtocells, between femto and outdoor macrocells.
  • Investigate how to effectively reduce power usage and radio wave pollutions in indoor environment.
  • Develop an automatic indoor radio network planning and optimisation tool suite that will fill the gap in the indoor RAN P&O market.
  • Facilitate knowledge integration and transfer between project partners, to enable cross-fertilisation between radio propagation modelling, wireless communications, operations research, computing, and software engineering.
  • Strengthen existing and create new strategic long-term collaborations between the participants and reduce the fragmentation in EU research in this vital area.

The objectives will be implemented by 126 PMs (Person Month) staff exchange and 120 PM new recruitments.

Achievement of these objectives will lead to the development of fast and accurate indoor radio propagation models and advanced indoor radio network P&O tools that can be used to plan and optimise WiFi, UMTS/HSPA/LTE, WiMAX for indoor coverage.

Work Package Work Package Title Lead Beneficiary Short Name Start Month End Month
WP-1 Indoor radio propagation modelling. INSA-Lyon 1 48
WP-2 The investigation of femtocells for indoor coverage. University of Bedfordshire 1 48
WP-3 The development of automatic indoor radio network planning and optimization tools. Ranplan 1 48
  Total: 3  

WP 1: Indoor radio propagation modelling


  • • Research and develop fast and accurate radio propagation models that can be used for indoor radio network planning and optimisation.

The WP contains the following R&D tasks.

T1.1 Computational efficiency

The first constraint our work should be aware of is the computational complexity at which target accuracy can be achieved. The discrete based approach from which our work will be derived is the Multi-Resolution Frequency Domain Partial Flows (MR-FDPF) method that was previously developed at INSA-LYON. This method exhibits very good properties for Indoor prediction but in turns may suffer from a high complexity in full 3D environments. To alleviate this limit, we will have to reduce the mathematical complexity of the method at the price, if needed, of a lower accuracy. In this project, the proposed approach relies on a multi-resolution decomposition of the problem that requires several matrix inversions. How to reduce computational complexity due to matrix inversion is critical. In this project, the reduction of the complexity will be achieved by subspace projection and singular value decomposition (SVD).

On one side these approximations reduce the memory, on the other side the SVDs computations and matrices reduction will require an increase of the processing time, hence they are computed only for the large MR-nodes matrices. We will focus in this work on the balance between accuracy and computational load.

T1.2 Statistical radio link prediction

Propagation simulators at a first glance are expected to provide the mean power. But fine simulations require also the prediction of the channel state including fading and impulse response of the channel, so that packet error rate statistics with a significant accuracy level can be predicted. This is indeed the exact input that is needed by network simulators to perform realistic simulations.

Many deterministic methods based on ray-tracing proposed to simulate the right impulse response. However, the lack of knowledge about the environment (furniture, people, materials) limits the accuracy of these approaches: the resulting predictions may be theoretically precise but not realistically accurate. Another efficient way consists of computing various statistical parameters. Obviously, a deterministic approach is used to predict large scale propagation phenomenon while small scale effects (fading, impulse response) are described in a statistical manner, thanks to a local statistical analysis of the simulated channel.

The fundamental idea holds on separation within the simulations of the reliable deterministic and the statistical parts which can be assessed only statistically. This work relies on experimental evaluation and an exhaustive statistical analysis of the simulations.

T1.3 Indoor-outdoor propagation predictions

As mentioned above, indoor to outdoor and outdoor to indoor propagation characterization is very important for a near future where several standards will have to share the same radio resource. LANs and MANs will coexist in the same band. In such a framework, it is expected that local equipment will contain a cognitive radio interface able to detect silent periods and allow transmission accordingly. Of course these resource sharing mechanisms may have a strong impact on the network performance.

In this framework, we propose to develop a full simulation framework by integrating our Indoor simulations into wide areas simulations. The interface between indoor and outdoor simulations will offer a good approach to perform these kinds of simulations.

T1.4 Measurement campaigns and model calibration,

The model predictions will be compared with real measurements. This task includes the following parts:

  • Measurement campaigns will be carried out to collect data.
  • Algorithms/methods will be developed to post-process measurement data.
  • Automatic model calibration algorithms.

WP 2: The investigation of femtocells for indoor coverage


  • Study interference between femtocells, between femto and outdoor macrocells.
  • Develop automatic configuration algorithms for femtocells.
  • Investigate handover strategies between femto and macrocells.

The details of the R&D tasks of this WP are as follows.

T2.1 The development of system level simulators

In this project, UMTS, WiMAX and LTE femtocells will be investigated. In order to carry out the above research & development tasks, system level simulators for UMTS, WiMAX and LTE need to be developed. CWiND (Centre for Wireless Network Design) at UoB has developed static and/or dynamic UMTS/HSPA and WiMAX simulators. In this project, an LTE simulator will be developed on the basis of the WiMAX simulator as they use similar technologies.

T2.2 The investigation of Interference due to femtocell deployment

Firstly, the interference between femtocells will be investigated. Two scenarios will be studied: first, dense residential areas such as terraced houses and multi-floored apartments, in which situations femtocells can be very close to each other; second, a multiple-floored office environment.

Secondly, the interference between femtocells and macrocells will be investigated. Typical femtocell deployment scenarios will be investigated.

The outdoor to indoor and indoor to outdoor radio propagation models developed in WP1 will be used to perform coverage and interference simulations that will be compared with real measurements (a spectrum analyser and antennas are necessary to perform this study).

The impact of femtocell interference to outdoor macrocells in terms of system capacity will be studied using the system level simulators developed in task T2.1.

T2.3 Auto-configuration of femtocells

Unlike macrocell deployment, femtocells are normally installed by the end user. Traditional cellular network planning approaches are not appropriate. It is not possible for mobile operators to plan the locations and parameters of femtocell APs. Therefore, a femtocell AP must be able to sense its radio environment and automatically configure its radio parameters so that its impact on the outdoor macrocells and neighbouring femtocells is minimized.

Automatic configuration will: select an available carrier with the least macrocell interference; tune the Tx power level of pilot signal; limit the maximum uplink and downlink transmit powers to suit the current radio conditions; monitor the radio conditions and automatically decide to reconfigure if the radio conditions change; raise an alarm to the femtocell management system if the level of interference prevents correct AP operation.

In this task, procedures and algorithms (e.g., how to decide Tx power) for femtocell auto-configuration will be developed.

T2.4 Handover strategies between femto and macrocells

When a UE moves into (out of) a home with femtocells, a handover between macrocell and femtocell will take place. During the handover from macrocell to femtocell: (re-)selection of the femtocell by the UE in idle mode; femtocell biasing parameters encourage switch from macrocell to femtocell with a lower quality value and receive level over the macrocell. During the handover from femtocell to macrocell: (re-)selection of the macrocell by the UE in idle mode; surrounding macro cell information is populated in the Home BS neighbour list; cell biasing discourages switching from femtocell to macrocell

In the project, the handover procedures and thresholds will be investigated so that the handover process can be optimized.

In the future, it is likely that femtocells will support different air interfaces, such as UMTS, WiMAX and LTE. In order to make best use of radio frequency, inter-system handover is very possible. Auto-selection of RATs and handovers are important issues to study towards developing cognitive radios. No open literature can be found in femtocell inter-system handovers.

WP 3: The development of automatic indoor radio network planning and optimisation tools


  • Develop an automatic indoor radio network planning and optimisation tool suite.
  • Integrate the work done at different partners.

The main R&D tasks in this WP are as follows.

T3.1 Indoor radio network P&O tool

A radio network P&O tool suitable for indoor environment will be developed. The indoor planning tool developed in this project will have an easy to use GUI. It will be able to import standard building data, e.g., formats like AutoCAD. The tool will allow user to modify/add obstacles and their constituting materials. The tool will include a library containing all the reference materials (concrete, plaster, glass…).

The tool will have APIs to link with radio propagation module and automatic RAN P&O module.

T3.2 Propagation model integration

The main input is the map of the environment and its properties. A material coefficients library as complete as possible must be developed. Some coefficients values can be found in the literature but probably some other measurements in different environments and frequency bands will need to be performed. Automatic import of building map and data is the best approach. But in some case where no data is available the best approach is to create the environment by hand with an easy GUI.

The second important data to take into account is the emitter. If a 3D model is used the full 3D antenna pattern will need to be simulated. But most of the antenna builders only provide 2D patterns (horizontal and vertical cut only). Models must be proposed to build the 3D antenna pattern. Some measurements are also necessary to validate the proposed models.

To have an optimization tool based on realistic signal coverage simulations, a calibration module must be included, to make the simulation fit the measurements. This calibration concerns both the environment data and the antenna data. The number of parameters to optimize can be very high, especially in complex buildings with a lot of different materials, and in the case of heterogeneous network with different kinds of emitters. Because operators want a quick calibration process during the optimization operation, research about complex function minimisation must be performed. Some approaches in the literature propose the use of usual Tabu or Simulated Annealing functions to perform this calibration. We think some new and more efficient methods can be developed, more adapted to our context.

T3.3 Automatic Network Planning and Optimization.

The first question is the definition, in mathematical terms, of the different properties that operators want to optimize. These mathematical functions will constitute the cost functions to minimize or maximize during the optimization.

The other point is the optimization method. Because, depending on the situation, different cost functions will have to be optimized. Moreover, these functions are very often concurrent. For example a solution that tries to increase the coverage will also increase the interferences, or a solution that decrease the price of the installation will decrease the total coverage. Thus a multi-objective approaches must be chosen.

This problem has been widely developed for outdoor propagation will GSM planning tools for example. But the Indoor resolution of such problems has not been investigated a lot and new cost functions and minimizations methods must be found. It is important to notice that, in multi-objective optimization, due to the high number of parameters and cost functions there is not one best solution but a high number of best solutions constituting the Pareto front of the resolution of the problem. An approach that would deliver different solutions depending of the priorities of the operator could be proposed, where some statistic and ranking of the solutions would help the operator making his final choice.

CITI Lab coverage

CITI Lab Coverage according to prediction

CITI Lab service zones

Serviced zones in CITI Lab according to prediction.