 Transmitters Separation for Single Frequency Networks Gilles BUREL and Pierre MAGNIEZ 1 L.E.S.T., Université de Bretagne Occidentale 6, Avenue Le Gorgeu - BP809 - 29285 BREST cedex - France tel : (33).2.98.01.62.46 fax : (33).2.98.01.63.95 email : Gilles.Burel@univ-brest.fr Abstract In a Single Frequency Network (SFN), the signals coming from nearby transmitters are mixed. Since the interfering signals can be seen as (very) long term echoes, an effective method to combat echoes, such as OFDM, is a natural candidate for SFN. However, although OFDM is a very efficient method, the very long delays that characterise SFN echoes require extremely long packets in OFDM transmissions and high stability of the subcarriers frequencies. An alternative approach based on array processing is proposed. The signals received by the sensors are filtered and downconverted. Then, source separation is performed using transmitters localisation and multiplication by the pseudo-inverse of the estimated mixture matrix. Symbol Timing and Carrier Recovery is then performed on each separated component. Furthermore, in order to increase the SNR, these components can be globally synchronised (using a correlator) and summed. Experimental results are finally presented and show good performances of the approach in a realistic configuration. Keywords Array Signal Processing in wireless systems, Signal Separation, Single Frequency Networks 1 1. Single Frequency Networks A Single Frequency Network (SFN) is a digital broadcasting network in which all the transmitters dedicated to the retransmission of a given program use the same frequency band. Hence, SFN requires much less spectrum than classical broadcasting networks because there is no frequency allocation. In a classical network, about 9 frequency bands per program are allocated: close transmitters use different bands in order to avoid interferences. SFN is believed to progressively replace classical broadcasting networks in the future. However, the price to pay for the gain on spectrum allocation is the need of more sophisticated signal processing. Since the transmitters use the same frequency band, the receiver gets a mixture of signals coming from the closest transmitters. Signal processing must be performed in order to separate the components of this mixture. 2. The traditional approach to SFN :OFDM Most research works about SFN consider the use of OFDM (Orthogonal Frequency Division Multiplexing). OFDM is indeed a good candidate for SFN, because of its power in reducing the effect of echoes. The basic idea is to consider the interfering signals as long term echoes. More precisely, the received signal in a given frequency band is a mixture of delayed versions of the basic signal. The delays correspond to the various propagation times (which are proportional to the distances between the transmitters and the receiver). As shown on figure 1, the signal received from transmitter 2 can be seen as a delayed version of the signal coming from transmitter 1. The delay is proportional to D 2 -D 1 (plus delays due to the paths from the original transmitter to transmitters 1 and 2). A typical propagation delay, for a difference of distances equal to 30 km, is 100Ps. With a symbol frequency equal to 10MHz, this corresponds to a delay of 1000 symbols. Hence, from the receiver side, SFN can be seen as a classical broadcasting network where strong echoes with huge delays are present. These “ echoes ” differ from short (IEEE Workshop on Signal Processing Advances in Wireless Communications, pp. 341-344, Annapolis, Maryland, May 9-12, 1999) ©1999 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.