The invention relates to a data transmission method utilizing combined CDMA/FDMA multiple access.

One major problem in the design and implementa¬ tion of mobile telephone systems is the efficient frequency utilization. As the amount of radio traffic increases continuously, the problem has become more and more important. The more efficiently the system is able to utilize an assigned frequency band, the greater is the possible number of system users. When new mobile telephone systems are developed, the radio channel multiple access methods are a central field of research.

The first generation of mobile telephone net¬ works is implemented by utilizing the FDMA method. The second generation that employs digital data trans¬ mission is designed on the basis of combined FDMA/TDMA technology.

The third radio channel multiple access method, CDMA, has been applied only recently on cellular net¬ works. CDMA is a multiple access method based on the spread spectrum technique, and it has several known advantages over the prior methods, e.g. it has no need for frequency planning, and it utilizes the frequency band efficiently.

In FDMA, users are distinguished from each other by means of frequency; each data signal of the user has a dedicated frequency band. In TDMA, the frequency band is divided into successive time slots, and the data signal of each user is transmitted in its own re¬ current time slot. In the case of combined FDMA/TDMA, several such frequency bands may be in use. In CDMA, the narrow-band data signal of the user

is modulated to a relatively broad band by a spreading code having a broader band than the data signal. Band widths such as 1.25 MHz, 10 MHz and 50 MHz have been used in known test systems. The spreading code con- sists of a number of bits. The bit rate of the spread¬ ing code is much higher than that of the data signal, and the bits of the spreading code are called chips to distinguish them from data bits and data symbols. Each data symbol of the user is multiplied by all of the chips of the spreading code. The narrow-band data signal thus spreads to the frequency band to be used. Each user has a separate code. Several users transmit simultaneously on the same frequency band, and the data signals are distinguished from each other in the receivers on the basis of the spreading code.

Correlators provided in the receivers are syn¬ chronized with a desired signal, which they recognize on the basis of the spreading code. The correlators restore the signal to the original narrow frequency band. On arriving at the receiver, signals modulated by another spreading code do not correlate in the receiver in an ideal case, but retain their wide band. One aims at selecting the spreading codes used by the system in such a way that they do not correlate with one another, i.e. are mutually orthogonal.

A typical feature of a cellular network environ¬ ment is that a signal propagating between a user and a base station does not propagate along a single straight path from the transmitter to the receiver but along several paths varying in length, depending on the properties of the environment. This kind of multi- path propagation occurs even though there would be direct visual communication between the base station and the mobile station. Multipath propagation is mainly due to the reflections of the signal from the

surrounding surfaces. Signals propagating along different paths have different transmission delays, and so they differ in phase on arriving at the receiver. Generally speaking, spreading codes are not orthogonal with all possible delay values. Signals with different delays therefore interfere with the detection of other signals. In other words, users interfere with one another, and this is called mul- tiple access interference. CDMA is an interference- limited system. The effect of multiple access inter¬ ference increases with the number of users, which degrades the signal-to-noise ratio of connections. With a certain number of users, the signal-to-noise ratio increases to such an extent that communication becomes difficult, and the number of users can thus be increased only at the expense of the quality of exist¬ ing connections. Typically, one radio channel having a bandwidth of e.g. 1.25 MHz may have simultaneously no more than 30 to 40 CDMA connections. If the channel capacity is to be increased at the base station, for instance, a new radio channel has to be taken into use. Even though the CDMA technique as such is more efficient than the prior techniques as far as the utilization of the frequency spectrum is concerned, the wide-band radio channels needed in the CDMA/FDMA system still occupy plenty of frequency band especial¬ ly within areas with a high capacity demand. Accord¬ ingly, there is a need for a still more efficient utilization of the frequency spectrum.

In a cellular network environment, users are positioned arbitrarily with respect to the base station and one another. The effect of multiple access interference is especially apparent in connections between mobile telephones and the base station. Tele-

phones close to the base station may block the trans¬ mission of more remote stations entirely, if the power control of the telephones is not accurate, as even a minor correlation of a strong signal may cause a major interference in the detection of a weak signal. This effect is called the near-far problem. The power control of telephones aims at ensuring that power received at the base station from each telephone would be equal irrespective of the distance between the telephone and the base station. However, accurate power control is difficult to realize due to e.g. the rapidly varying character of the radio channel.

Problems caused by multiple access interference could be dispensed with if the users' signals would be fully orthogonal, i.e. if they would not correlate. Therefore attempts have been made to create code families in which cross-correlation between the codes is minimized.

The object of the present invention is to provide a data transmission method, where multiple access interference and problems caused by it can be reduced substantially by utilizing certain type of codes and their properties such that a higher capacity and spectral efficiency can be achieved in the data transmission system.

This is achieved by a method according to the invention, which is characterized in that comple¬ mentary codes are used as CDMA spreading codes; that interlaced frequencies are used at least at some frequency bands, the frequency difference between the interlaced frequencies being selected from the zero points of auto- and cross-correlation functions cal¬ culated as a function of the frequency offset of the CDMA spreading codes used on the frequency channels; and that transmissions on the frequency channels used

are synchronized with one another.

The invention also relates to a CDMA/FDMA radio system having a plurality of frequency channels. The system is characterized in that the CDMA spreading codes used in the system are complementary codes, and that at least some of the frequency channels are interlaced frequencies having a mutual frequency difference equal to the frequency difference between the zero points of auto- and cross-correlation functions calculated as a function of the frequency offset of the CDMA spreading codes used on the frequency channels, and that transmissions on the interlaced frequency channels are synchronized with one another. The invention utilizes a spreading code family having ideal cross-correlation properties when certain marginal conditions are met. This kind of code family is called complementary codes. A characteristic feature of complementary codes is that they consist of a plurality of members of equal length so that the sum of cross-correlations between these codes formed by combining is zero with all delay values. In other words, complementary codes are fully orthogonal with respect to one another. However, all of the members that constitute the complementary codes used as spreading codes are not as such mutually orthogonal with all delay values. The members of the complemen¬ tary code therefore have to be transmitted in mutually uncorrelated channels in order that interference could be avoided. In practice, this can be realized in many different ways. One alternative is to transmit the code members at different frequencies. The spacing between the frequency bands thereby should be at least equal to the coherence bandwidth. In the case of CDMA, it is more advisable to separate the code members from

each other in the time domain. A sufficient guard time covering timing inaccuracy, delay spread and the effect of propagation delays should be left between the members. Other ways of realizing uncorrelated channels include the use of different polarization levels or quadrature components of the carrier.

One major advantage of the invention is that when complementary codes are used, it is possible to transmit signals of several users on the same fre- quency band without multiple access interference. This is possible if all of the users transmit in synchron¬ ization, whereby the different members of the comple¬ mentary codes will not overlap during transmission and correlate. By examining the mutual cross-correlation functions of the codes at transmitting frequency off¬ sets smaller than the frequency spectrum required by the code, it is possible to find frequencies such that signals transmitted at them do not at all correlate with each other and thus will not interfere with each other. The radio channels of the CDMA/FDMA system can be selected so that the frequency difference between the carriers of adjacent radio channels is equal to the frequency difference between two such zero points, i.e. smaller than the bandwidth required by the spreading code used, whereby the bands of adjacent radio channels overlap. This "interlacing technique" allows a considerably more efficient use of the frequency spectrum than in the prior CDMA/FDMA systems. In the following the invention will be described more fully with reference to the attached drawings, where

Figure 1 shows an example with two users; Figures 2a to 2d show an example of a system of two users;

Figure 3 shows an example of the spreading of the data signal of one user;

Figures 4a and 4b show inaccuracy functions for complementary codes; Figure 5a illustrates frequency band allocation in a conventional CDMA system;

Figure 5b shows an example of the band alloca¬ tion of interlaced frequencies;

Figure 6 illustrates by way of example the implementation of a receiver operating by the method according to the invention; and

Figures 7a and 7b illustrate by way of example the implementation of a transmitter operating by the method according to the invention. The present invention utilizes complementary codes as spreading codes. Complementary codes are described more closely in a doctoral thesis by B. P. Schweitzer, Generalized Complementary Code Sets , 1971, University of California, Los Angeles, USA, 87 pages. In the data transmission system each user must have a separate spreading code. If the capacity of the system is K users, K orthogonal codes are needed, which can be realized as complementary codes each consisting of K members. K uncorrelated channels are required for the transmission of each code member.

Figure 1 shows by way of example a system with two users. Each user has a separate code with two members 14a, 14b and 15a, 15b, respectively. The code members are transmitted in uncorrelated transmission channels 10 and 11, which may be implemented e.g. as different frequencies or time slots. At the reception end there are provided filters 16a, 16b and 17a, 17b, respectively, matched with the codes used, and the users' data signals S18 and S19 are obtained by combining the output signals of the filters.

The signals of the two-user system shown in Figure 1 will be described by way of example below. Assume that the system employs BPSK modulation. Modu¬ lation states used are indicated by "+" and "-" signs. Assume that the following two mutually orthogonal complementary code pairs are used in the system: s _1:L =+++-, s ₁₂ =++-+, and s ₂₁ =+-++ and s ₂₂ =+ . Of these s _n and s ₁₂ are assigned to user 1, and s ₂₁ and s ₂₂ to user 2. At the reception the impulse responses of the matched filters are thus: filter 16a of user 1: -+++, and filter 16b: +-++; filter 17a of user 2: ++-+, and filter 17b: +. If it is assumed that the users' data signals are binary, d _x = 1 and d ₂ = 0, the wave¬ forms of user 1 are +++- at the output of 14a, and ++-+ at the output of 14b. Correspondingly, the wave¬ forms of user 2 are +-++ at the output of 15a and + at the output of 15b.

Figures 2a to 2d show signals in the receiver when it is assumed that the radio channel is ideal, i.e. no interference and noise occur. Gain is assumed to be 1. Figure 2a shows the signals at the outputs of the matched filters 16a and 16b, and a summed signal 18 when user 1 has transmitted. Correspondingly, Figure 2b shows the signals at the outputs of the matched filters 17a and 17b, and a summed signal 19, when user 2 has transmitted. A waveform similar to an ideal impulse response is obtained for each user, from which waveform the transmitted data symbol can be detected. Figure 2c shows the signals at the outputs of the matched filters 16a and 16b and a summed signal S18 when user 2 has transmitted. Correspondingly, Figure 2d shows the signals at the outputs of the matched filters 17a and 17b, and a summed signal S19 when user 1 has transmitted. No waveform is detected in either one of the summed signals, that is, no

multiple access interference occurs. This is due to the orthogonality of the spreading codes.

In the preferred embodiment of the invention, the transmission of the data signal of one user takes place as shown in the example of Figure 3. Assume that the system uses a complementary code family consisting of K members SI, S2,..., SK. Each member is N chips in length. The duration of each chip is T _c seconds. Ac¬ cordingly, the format of the code of user i is S _j l, S _± 2,..., S _j k, where Sik (k = 1, 2,...,K) is one of the members SI, S2,..., SK, so that each code is ortho¬ gonal with respect to the other codes.

Figure 3 illustrates the transmission of one signal burst, where n data bits or data symbols b _n are transmitted. At the beginning of the burst each data symbol b _n is modulated by the first member S^ of the spreading code. The duration of this transmission is nNT _c seconds. Then a transmission break follows. The purpose of the transmission break is to separate successive code members from each other in the time domain. Even though the complementary codes are mutually orthogonal, the code members as such are not uncorrelated with all delay values. Therefore the duration of the transmission break has to be suf- ficient, in order that cross-correlation between the code members would not appear in the receiver due to multipath propagation taking place during the trans¬ mission and timing inaccuracies. A required duration of the transmission break can be defined as T _M + T _R + T _A seconds, where T _M is the multipath spread, T _R is the difference between signal propagation delays, and T _Δ is the timing inaccuracy. After the transmission break each data symbol b _n is modulated by the following member S^ of the spreading code, whereafter a similar transmission break again follows. In this way, the

data symbols are modulated one after another by all members of the spreading code.

The invention thus utilizes a combined CDMA/FDMA method. A plurality of frequency bands are used, and the users are distinguished from each other on each frequency band on the basis of the spreading code. Combining CDMA and FDMA as such is previously known, but the way in which the frequency bands to be used are selected in the solution according to the inven- tion is new as compared with the prior methods and considerably more efficient in terms of frequency efficiency. The selection is based on the zero points of the cross-correlation functions of complementary codes calculated at the transmitting frequency off- sets.

Figure 4a shows by way of example auto- and cross-inaccuracy functions calculated for the comple¬ mentary codes. The inaccuracy function describes auto- and cross-correlation when the codes have a mutual frequency offset. In this specific example, each code has four members. Each code member consists of four chips. The spreading ratio is thus sixteen. In Figure 4a, the first curve drawn by a continuous line re¬ presents the autoinaccuracy function of the code. The horizontal axis has been scaled as follows: 1.0 stands for a phase change of 2*τt during a chip, that is, 0.25, which is the maximum point of the horizontal axis stands for a phase shift of π/2, i.e. 90 degrees, due to a frequency error. In the same figure, the corresponding cross-inaccuracy functions are drawn by a broken line. It is to be seen that three common zero points can be found for the curves in this specific case. Frequency offsets corresponding to these zero points can be used as channel spacings in the system, that is, the frequencies indicated by the zero points

can be used as frequency channels (carriers) in the system. These frequencies are called interlaced fre¬ quencies.

Figure 4b shows the same curves as Figure 4a, but the maximum value of the horizontal axis has now been increased up to 1.0. It can be seen that com¬ ponents deviating from zero occur at intervals of τι/2, which in this specific case prevents the use of more than four adjacent interlaced frequencies. The follow- ing interlaced frequencies cannot be used until after the transmitting filters of the transmitters separate the frequency bands from each other, i.e. after 2*τt. This corresponds to the normal difference between FDMA frequency bands. In the example, one bit is thus transmitted with a spreading of four chips, and as shown in Figure 4a, there are four interlaced frequencies available. The channel spacing, i.e. the difference between the frequencies, is the bit frequency. The frequency band of each channel is, however, much greater than the bit frequency, and so the spectra of the channels overlap almost completely. If the spreading ratio, which was four in the example described above, is increased, the number of available interlaced frequency channels increases correspondingly, whereas the channel spacing remains unchanged.

Figure 5a illustrates a conventional method for allocating different frequency channels in the CDMA system. In the example shown in the figure, three frequency ranges 50, 51 and 52 having center fre¬ quencies f _a , f _b and f _c , respectively, are used. Accord¬ ing to the FDMA principle, in order that transmissions at the different frequencies would not interfere with one another, the spectra at the different center fre- quencies must not overlap.

Figure 5b illustrate the method according to the invention for the allocation of frequency channels and available frequencies in proportion to the width of the spectrum. The figure is intended only to illus- trate the method; it is not a genuine pattern produced by a spectral analyzer. The spectrum of a transmission taking place at a frequency f ₀ has the width of area 53. The method according to the invention also allows interlaced frequencies f _{l t} f ₂ and f ₃ to be utilized. Corresponding spectra are drawn as areas 54, 55 and 56, respectively. It can be seen that the spectra of the signals at different interlaced frequencies over¬ lap almost completely, but as the carrier frequencies are selected on the basis of the zero points of the spreading code inaccuracy functions, the frequencies will not interfere with one another. The following cluster of interlaced carrier frequencies is at frequencies f ₄ , f ₅ , f ₆ and f ₇ , and the corresponding spectra are 57, 58, 59 and 60. As compared with Figure 5a, it is noted that the method according to the in¬ vention enables a more efficient frequency util¬ ization.

Interlaced frequencies can be used within a single cell or they can be allocated to different cells. As the network is synchronized, there does not either occur intercell multiple access interference between the interlaced carrier frequencies, provided that the factor T _R in the transmission break as defined between the members of the spreading code also covers the difference between propagation delays between cells. In practice, this limits the size of system cells to some extent, as it is not advisable to have a transmission break of unlimited length.

The same frequencies may also be re-used in different cells, provided that these cells are suffi-

ciently remote from each other so that interference will not occur on the same frequency. In FDMA this is realized by conventional frequency planning, whereby the number of frequencies corresponds to the re-use cluster. In principle, interference originating from outside the cluster occurs to some extent, but due to the long propagation delay the orthogonality condition is not met, and the interference appears as normal CDMA multiple access interference, that is, is close to white noise.

Conventional CDMA uses processing gain to combat multiple access interference. In the method according to the invention, multiple access interference occurs only in interference originating from outside the re- use cluster. In theory, processing gain is thus needed only to combat thermal noise and the above-mentioned interference. On the other hand, the propagation delay spread requirement, i.e. the cell size limit, can be moderated if the system capacity is good e.g. due to high processing gain.

Figure 6 illustrates a simplified configuration of a base station receiver operating by the method according to the invention. For the sake of clarity, the figure shows only two receiver units 61a and 61b. In reality, the number of base station receivers is much greater. The receiver typically comprises a filter 62a, 62b, a first multiplier 63a, 63b, a second multiplier 64a, 64b, a radio-frequency oscillator OSCl, 0SC2, a demodulator 66a, 66b, and means 65a, 65b for generating a spreading code. For the sake of clarity, the receivers shown in the figures comprise only one spreading code demodulator unit 64-66. In reality, the CDMA receiver generally comprises several such spreading code demodulator units for each con- nection (CDMA channel), which allows several multi-

path-propagated components to be received and summed in a manner typical of CDMA. In addition, several CDMA channels are typically transmitted on the same carrier, each channel having a dedicated group of spreading code demodulator units, to which an output signal from a multiplier 63 common to all of the demodulators is applied.

Figures 7a and 7b illustrate a simplified configuration of two transmitters operating by the method according to the invention. The transmitter typically comprises a filter 72a, 72b, a first multi¬ plier 73a, 73b, a second multiplier 74a, 74b, a radio- frequency oscillator OSCl', OSC2' , and means 75a, 75b for generating a spreading code. The transmitter also comprises a dedicated multiplier 74 and spreading code unit 75 for each CDMA channel to be transmitted on the same carrier, the output signal from the multiplier and the spreading code unit being applied to a common multiplier 73. Assume that the receiver 61a receives a signal from the mobile station transmitter 71a, and the re¬ ceiver 61b receives a signal from the mobile station transmitter 71b. The user's data signal 76a, 76b arrives at the transmitter 71a, 71b, and it is multiplied by the user's spreading code in the multi¬ plier 74a, 74b. The output signal of the multiplier 74a, 74b is modulated by a radio-frequency signal obtained from the oscillator OSCl' , OSC2' in the multiplier 73a, 73b. This modulated signal is filtered in the passband filter 72a, 72b and then applied to an antenna 70a, 70b. The radio frequencies of the oscil¬ lators OSCl, 0SC2 are selected so that they coincide with the zero point of the cross-inaccuracy functions of the spreading codes used, which is a characteristic feature of the invention. By way of example, assume

that each one of the above-mentioned connections is at a different frequency. In a real situation, one fre¬ quency is naturally assigned to several users. The transmitting frequency of OSCl' is indicated with f ₀ , and the transmitting frequency of OSC2' with f ₁ . According to the invention, the frequencies of OSCl' and OSC2' deviate from each other by the bit fre¬ quency, and their spectra overlap almost completely, as shown in Figure 5b. Similarly, the signal received at the receivers 61a and 61b is filtered in the bandpass filters and then first multiplied by a radio-frequency signal obtained from the oscillator OSC'1, 0SC2 and then by the user's spreading code in the multiplier 63a, 63b, whereafter the data can be demodulated. The fre¬ quencies of the oscillators OSCl and OSC2 deviate from each other by the bit frequency, as described above.

The attached figures and the description related to them are only intended to illustrate the present invention. In this details, the method according to the invention may vary within the scope of the attached claims.