Advanced NeTwork radio Identification equipment for Universal Mobile communications

Operators of UMTS/FDD cellular communications systems will require efficient network planning to distribute the valuable available resources or frequencies. To do so, the ability to identify node-Bs in the field is necessary. A classical trace mobile does not have sufficient sensitivity or signal separation capacity to identify low interfering signals. Space-time processing based on smart antenna techniques is used to suppress co-channel interference caused by other base stations and significantly improve sensitivity. Identification of the interfering node-Bs first includes the synchronization with all the surrounding base stations and then the demodulation and the interpretation of the Primary Common Control Physical Channel (P-CCPCH) providing valuable information such as the Cell Identity (CI), the Mobile Country Code (MCC) or the Mobile Network Code (MNC). Since UMTS-FDD mode provides synchronization channels containing a known primary synchronization, we can use a detection approach to obtain synchronization. The aim of such a method is to compute a decision statistic for every possible time instant, which can be processed in further stages to decide whether or not a synchronisation sequence is present. The initial objective of ANTIUM was to detect BTS having a P-CPICH Ec/I0 as low as -25dB, since it was shown that such BTS could have an impact on the mobile performance when they are fully loaded. If we consider P-CPICH Ec/I0 between -20dB and -15dB, a mono-channel processing allows detecting only about 61% of the BTS detected with a 5-channel one. This percentage drops to 17% if we consider P-CPICH Ec/I0 between -25dB and -20dB. Therefore a multi-channel processing improves significantly the detection performance and is required to reach the ANTIUM objectives. A 3 or a 4-channel ANTIUM equipment seems to be a good trade-off between performance / cost / complexity, since it allows to detect respectively 64% and 78% of the BTS detected with 5 channels in the range [-25dB; -20dB]. For identification, several multi-sensor demodulation algorithms were developed. As the spreading factors and the spreading codes allocated to the channels transmitted by the detected node-Bs are unknown, it is impossible to use joint detection algorithms in order to detect the symbols transmitted by the P-CCPCH. We therefore study a family of sub-optimal receivers, which only require the knowledge of the spreading code allocated to the P-CCPCH. Roughly speaking, each of them consists in estimating the chip sequence by a spatio-temporal filter. This filter can be interpreted as a multi-channel equalizer because, ideally, it allows both to compensate the effect of the propagation and to cancel signals due to interfering node-Bs. The studied algorithms are the following ones: 2D Rake, Spatial Wiener Filter, Spatial Wiener Filter with Filtered Reference, Spatial Rake with Spatial Whitening, Space-Time Wiener Filter. The Spatial Rake with Spatial Whitening (SRSW) presents the best trade-off in terms of complexity / performance. With a 5-channel processing, the SRSW allows to demodulate the BCH information with a good CRC rate better than 90% when the P-CCPCH Eb/I0 is greater than -1dB (which corresponds to a P-CPICH Ec/I0 greater than -20dB if the P-CCPCH level is 2dB below the P-CPICH one). In comparison, the loss of performance by using a 1-channel processing is equal to 11dB. Moreover, the UMTS-FDD standard foresees the use of transmit diversity on different downlink physical channel types. Therefore, all the developed algorithms were extended to the Space-Time Transmit Diversity mode that is used on the P-CCPCH. In a first stage, the algorithms have been developed and tested in simulation using a C environment. The results have been published in several conference papers and a doctoral dissertation. Then they have been implemented on a demonstrator and validated on the field on real UMTS/FDD networks. The performance describe above comes from these field trial measurements. The algorithmic study is detailed in D3011b and the validation on the field in D5011.

This result was obtained within the algorithmic study for UMTS/TDD (WP3.2). Signal processing algorithms for a network-monitoring device have been developed that enables operators of UMTS/TDD cellular communications systems to monitor and improve their network. The basis for these improvements in network quality is interference analysis. The use of multiple antennas and sophisticated multi-user space-time signal processing algorithms allows estimating the interference levels of many surrounding base stations (including base stations with weak power). The signal processing algorithms concern the stages of synchronisation, channel estimation, and data detection. The overall goal is to demodulate the system information transmitted on the broadcast channels of the different base stations. The key innovation of this result was to adapt and extend advanced signal processing techniques (e.g. MMSE filtering, GLRT hypothesis testing, DFB detection) to the special demands of network monitoring in UMTS/TDD, exploiting the availability of multiple receive antennas in an off-line signal processing mode. The algorithms have been developed and tested in simulations using the simulation environment Matlab. The results have been published in several conference papers, a journal paper (currently in the review process), and a doctoral dissertation. Within the ANTIUM project, a demonstrator of the network-monitoring device has been developed for DVB-T and the FDD mode of UMTS. In order to include the TDD mode of UMTS in this demonstrator, the further development steps of C-code development and demonstrator integration have to be carried out.

For a UMTS operator, the ANTIUM tool is very useful in order to analyse interference situations, to optimise the radio planning process and also to estimate the cellular capacity from simple measurements. In fact, by measuring different radio parameters such as P PCICH level, Ec/I0, total interference factor, number of BTS in the active set, it is not only possible to optimise the 3G network planning process but it is also possible to analyse and estimate the network cellular capacity. Combining ANTIUM field trials measurements and a simulation tool developed for that purpose, it allows us to estimate the capacity of a site or the impact of certain radio planning parameters such as the active set margin on capacity and power consumption. This is very valuable for an operator during the deployment phase and afterwards. For example, it is possible to use estimated traffic volume as an input and estimate the needed BTS power in order to decide whether the network planning is correct or if it will be necessary to add new base stations in the future (as a function of traffic volume forecast). Some tests based on a fiels trial in Paris have been performed and used as an input for the capacity estimator. One of the conclusions we can draw is that intermediate values of the active set margin (3dB or 5dB) are more suitable for resource consumption optimisation. Thus, by reducing the required base station power, soft handover optimisation can increase capacity by a factor of 30%. We can see that this effect depends on the environment (different optimal soft handover margins for different sites). This result is non-trivial and very valuable for an operator. Another possible operational use for a UMTS operator is the analysis of the impact of the target Block Error Rate (BLER). This is an important parameter in the QoS profile that characterizes the UMTS Radio Bearer. It is usually set to 1% for conversational services such as voice and to 10% for data services such as web browsing (because these services often rely on layer 2 retransmissions). Of course, when the target BLER decreases, the power required serving users increases. These results also help the operator to tune the QoS profile parameters as a function of the network performance. Given a certain configuration, (active set margin, traffic model, BLER, site...), it is also possible to determine the number of Erlang a cell is able to serve for a given blocking rate. Moreover, such a graph can be established for any service (voice and data) at any bit rate.

Providing the operators with a measure of the quality of the downlink transmissions is very useful, because it gives a good picture of the impact of the network deployment on the performance. As mobile receivers will probably use conventional RAKE receivers in order to demodulate the transmitted symbols, we propose to estimate from the received signals the signal to interference plus noise ratio (SINR) at the output of such a receiver. To do so, we showed that the SINR can be expressed in closed form in terms of the channel impulse responses between the active base stations and the receiver, and from the total power transmitted by each active base station. Using high quality CIR estimation schemes thus allows evaluating the SINR, and thus the corresponding bit error rate. One of the interest is that from measurements performed at one given time with a given network load, the formula allows to extrapolate the obtained BER for different loads of each detected BTS and thus to deduce capacity estimation. Moreover, one of the originality of the ANTIUM tool is that it provides the operator with different valuable factors that other classical tools do not produce. These factors, directly extracted from the SINR formula are the orthogonality loss factor (characterising the intra cell interference due to multi-path propagation) and the total interference factor (characterising both intra-cell and extra-cell interference). Operators commonly use the Ec/I0 as a criterion for network planning. The total interference factor appeared to be more efficient criterion that is directly linked with a mobile performance. This factor does not depend on the traffic load of the network at the measurement moment and it accurately takes into account the propagation characteristics, the mobile Rake processing and soft hand-over configurations.

During the deployment of a new DVB-T network, the network operator has to check that it does not generate interference on the existing networks. If it is the case, it must identify the transmitter that generates this interference. This identification issue will become more difficult and critical with the densification of the DVB-T networks since on the one hand the number of DVB-T signals that may interfere will increase, and on the other hand, when we consider a portable reception, there is a multi path environment with no direct path from the transmitter to the receiver. When we consider a classical configuration for the DVB-T signal such as the 8K mode, a guard interval of 1/32 and a 64 QAM constellation with a code rate of 2/3, the objective is to detect and to identify all the DVB-T signals that are in a given channel for C/I down to -20dB. To reach this performance, a multi-antenna receiver is mandatory and space-time algorithms have been developed in the field of the ANTIUM project: - The first step of the ANTIUM receiver consists in detecting all the DVB-T signals that are present in a given channel, and in doing the coarse-time synchronization, which corresponds to find for each of them the beginning of the useful part of the OFDM symbols. This is achieved by doing at each instant the correlation between a reference sequence and the output of the spatial filter matched to this reference sequence. When this statistics is higher than a pre-determined threshold, the detection is effective. Moreover, the reference sequence we have chosen also provides the location of the scattered pilots. - The second step consists in doing for each detected signal the OFDM demodulation (FFT of the useful part of the OFDM symbols) on each antenna and in doing the channel estimation. A new channel estimator, able to deal with co-channel interference, has been designed. After a preliminary rough estimation, at the scattered pilot location, of the channel in the time-frequency domain, the estimation is transformed to the time-delay domain where a parametric model is considered. The estimation is then improved using a MMSE algorithm based on the estimated statistics of the channel. This channel estimation allows providing an estimation of the level of the considered DVB-T signal as well as its SINR. - The last step consists in demodulating the TPS bits for each detected DVB-T signal. The TPS provides some information and more specifically the Cell_id, which is an identifier of the cell to which the transmitter belongs. To achieve a reliable TPS demodulation even for C/I down to -20dB, MMSE space-time combining is used and we take profit from the frequency diversity of the TPS. All these algorithms are detailed in the D3031b deliverable. They have been implemented in the ANTIUM demonstrator, and their performance have been assessed not only by simulation, but also with lab tests and during field trials whose results are detailed in the D5021 deliverable.

Multi channel interference analysis systems have conventionally used high-performance high-priced receivers. The development of a low cost multi channel receiver has served to effectively prove the feasibility of the development of a dedicated UMTS band low cost unit. During the project, the impact that different architecture issues and technologies have in receiver performance and cost has been studied. The conclusion is that a conventional super heterodyne architecture, with low cost commercial electronics, and without internal calibration system may obtain similar results to a high performance receiver at a small fraction of the cost only. Special attention needs to be paid to local oscillator coherency and to intermediate filter selection, since they represent the main source of errors between channels. A lesson has been learned about maximum gain and needed dynamic range of the receiver for a multi channel tool due to received signal characteristics measured in different environments, which have been established in 60dB maximum gain and 50dB dynamic range. The knowledge gained in receiver development will be used in different ongoing and future in-house Telefonica system R&D within TID, all of which integrate different kinds of UMTS receiver. An important part of the systems will be industrialised by different manufacturers, and then installed a used in current 3GPP networks.

Risultati finali

Condividi questa pagina

Scarica