A transmitter, receiver and methods for transmission / receipt of a communication signal based on overlapped orthogonal chirp waveforms

Abstract

The present disclosure relates to a transmitter arranged to transmit a communication signal x comprising batches of data, each batch comprising N transmit amplitudes, an, n=1, 2, N, representing information to be transmitted. The transmitter comprising a modulator arranged to modulate the communication signal x from superposition of N a linear frequency sweeps, each frequency sweep having an assigned transmit amplitude, an, of the batch representing information carried by the particular frequency sweep, a predetermined frequency range B common for the N frequency sweeps, and a predetermined sweep time T common for the N frequency sweeps. The time difference between start of consecutive sweeps is the inverse of the frequency range B. The sweep time T is larger than the inverse of the frequency range B. The present disclosure also relates to a corresponding receiver for receipt of a signal y transmitted through a communication channel.

Description

[0001] A transmitter, receiver and methods for transmission / receipt of a communication signal to avoid subcarrier fading

[0002] TECHNICAL FIELD

[0003] The present disclosure relates to a transmitter and method for transmission of a communication signal comprising batches of data, each batch comprising N transmit amplitudes, an, n=1, 2,..., N, representing information to be transmitted.

[0004] The present disclosure further relates to a receiver and method for receipt of a signal y transmitted through a communication channel.

[0005] BACKGROUND

[0006] In telecommunications, orthogonal frequency-division multiplexing, OFDM, is often used in wideband digital communication in modulation for encoding data on multiple carrier frequencies.

[0007] OFDM offers an attractive way of multipath handling, i.e. setting up the conditions required for solving

[0008] xRX (0=Jz{t^r}xTX (j- ~

[0009]

[0010] 0

[0011] for wide band communication, wherein in communication the convolvant

[0012]

[0013] provides the attenuation at time t of the variously delayed propagation paths, each path originating from the transmitted signal a time t- r. The convolvant

[0014]

[0015] z(t,r) is referred to as the transfer function. OFDM relies on that the transfer function can be considered shift-invariant. Thereby, to a certain extent OFDM is insensitive to propagation path changes, i.e. Doppler effects. However, in the course of time the accumulating change in

[0016]

[0017] z(t,r) requires a recalibration thus obtaining an updated transfer function. The expected degree of shift invariance determines the length of time between recalibrations. OFDM divides available bandwidth into many narrow subcarriers. Each subcarrier carries part of the transmitted data in parallel. Due to multipath propagation and Doppler effects, some subcarriers may experience deep fades (signal attenuation), where the power of the signal in specific subcarriers drops significantly. This can result in data loss or errors on those subcarriers.

[0018] Orthogonal Time Frequency Space, OTFS, another modulation scheme used for transmission of symbols.

[0019] SUMMARY

[0020] It is an object of the present disclosure to mitigate, alleviate or eliminate one or more of deficiencies and disadvantages in the prior art including impairments of subcarrier fading, which appears the foremost deficiency in both OFDM and OTFS.

[0021] The present disclosure relates to a transmitter arranged to transmit a communication signal x comprising batches of data each batch comprising N transmit amplitudes, an, n=1, 2,..., N, representing information to be transmitted. The transmitter comprises a modulator arranged to modulate the communication signal x from superposition of N linear frequency sweeps. Each frequency sweep has an assigned transmit amplitude, an, representing information carried by the particular frequency sweep. Each frequency sweep has a predetermined frequency range B common for the N frequency sweeps. Each frequency sweep has a predetermined sweep time T common for the N frequency sweeps. The time difference between start of consecutive sweeps is the inverse of the frequency range B. The sweep time T is larger than the inverse of the frequency range B.

[0022] The communication modulation technique used circumvents the impairments of subcarrier fading, which as stated above appears the foremost deficiency in both OFDM and OTFS In accordance with the present solution, all signal components have the full frequency range B. This is in contrast to OFDM, wherein the signal components are the subcarriers with bandwidth equal to the reciprocal of symbol time.. Therefore, the problem arising from subcarriers experiencing deep fades (signal attenuation), where the power of the signal in specific subcarriers drops significantly, does not occur with the present solution. Due to fading there can never be any certainty that an OFDM signal will be fully retrieved. Typically, in the various OFDM arrangements proposed, an invariable figure of merit is the resulting bit error rate - BER, thus acknowledging to that these schemes cannot be expected to operate error-free. This problem is removed or at least alleviated with the solution as presented therein, as the solution as presented herein avoids subcarrier fading.

[0023] In detail, using the modulation as presented herein, fading will impact equally the set of signals representing the amplitudes to be communicated. As the communication band fluctuates between frequency intervals of constructive and destructive interference, transmission will take advantage of those intervals where fading conditions are favourable. The size of the batches of data each comprising N transmit amplitudes may be limited by acceptable delay, and / or storage space and / or need for-recalibration.

[0024] Embodiments of the invention are given in the dependent claims.

[0025] The present disclosure further relates to a receiver for receipt of a signal / transmitted through a communication channel, said receiver comprising

[0026] a linear frequency sweep generator arranged to generate consecutive linear frequency sweeps Ym; m=1,2,... having a common amplitude. Each frequency sweep hasa predetermined frequency range B common for the frequency sweeps and corresponding to the frequency range of a transmitted signal, and a predetermined sweep time T common for the frequency sweeps and corresponding to the sweep time of the transmitted signal, wherein the sweep time T is larger than the inverse of the frequency range B,

[0027] a data buffer arranged to simultaneously store at least T*B sweeps and to thereby store the generated linear frequency sweeps during their duration, and

[0028] a demodulator arranged to sample the signal y received at the receiver to provide sample values y(tn), and to modify each sweep in the data buffer through multiplication with the sample values y(tn) to provide a product y(tn)Ym(tn), and wherein a summation is made for each sample time tnto obtain at least T*B demodulated amplitudes bm. The consecutive linear frequency sweeps Ymconstitute the complex conjugates of Xn, with the same sweep gradient and sweep time.

[0029] Each such sweep is stored in the data buffer for its duration in the receiver. At each time tn, the signal y received from the transmitter is sampled, which gives the sample values y(tn). Each sweep in the data buffer is modified by these sample values y(tn) by multiplication Ym(tn) -> y(tn)Ym(tn). As discussed above, as a final demodulation step, summation is performed over each sweep.

[0030] The demodulated amplitudes bmare thus provided as

[0031] bm= Sn{y(tn)Ym(tn); |tm - tn| < T}.

[0032] Embodiments are given in the dependent claims.

[0033] The present disclosure further relates to a method performed ata transmitter for transmission of a communication signal x comprising batches of data, each batch comprising N transmit amplitudes, an, n=1, 2,..., N, representing information to be transmitted, said method comprising

[0034] modulation of the communication signal x by superposition of N linear frequency sweeps, each frequency sweep having

[0035] an assigned transmit amplitude, an, representing information carried by the particular frequency sweep,

[0036] a predetermined frequency range B common for the N frequency sweeps, and a predetermined sweep time T common for the N frequency sweeps,

[0037] wherein the time difference between start of consecutive sweeps is the inverse of the frequency range B, and

[0038] wherein the sweep time T is larger than the inverse of the frequency range B.

[0039] The present disclosure further relates to a method performed at a receiver for processing of a received signal / transmitted through a communication channel, said method comprising generating consecutive linear frequency sweeps Ym; m=1,2,... having a common amplitude, each frequency sweep having

[0040] a predetermined frequency range B common for the frequency sweeps and corresponding to the frequency range of a transmitted signal, and a predetermined sweep time T common for the frequency sweeps and corresponding to the sweep time of the transmitted signal,

[0041] wherein the sweep time T is larger than the inverse of the frequency range B,

[0042] storing the generated linear frequency sweeps during a time period long enough to simultaneously store at least T*B sweeps,

[0043] sampling the received signal y to provide sample values y(tn),

[0044] modifying the stored linear frequency sweeps through multiplication with the sample values y(tn) to provide a product y(tn)Ym(tn), and

[0045] forming a sum for each sample time tnto obtain at least T*B demodulated amplitudes bm- At least some of the embodiments as defined herein may be especially suited for use by objects in motion, for example vehicles such as high speed trains.

[0046] It is to be understood that the herein disclosed disclosure is not limited to the particular component parts of the device described or steps of the methods described since such device and method may vary. It is also to be understood that the terminology used herein is for purpose of describing particular embodiments only, and is not intended to be limiting. It should be noted that, as used in the specification and the appended claim, the articles "a", "an", "the", and "said" are intended to mean that there are one or more of the elements unless the context explicitly dictates otherwise. Thus, for example, reference to "a unit" or "the unit" may include several devices, and the like. Furthermore, the words "comprising", "including", "containing" and similar wordings does not exclude other elements or steps. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The above objects, as well as additional objects, features and advantages of the present disclosure, will be more fully appreciated by reference to the following illustrative and nonlimiting detailed description of example embodiments of the present disclosure, when taken in conjunction with the accompanying drawings.

[0048] Figure 1 is a schematic overview illustrating the principle of transmit signal generation in an example of a transmitter

[0049] Figure 2 is a schematic illustration of N transmit amplitudes (an, n=1, 2.... N) of the transmitter above designed to make the respective batch of data partly cyclic.

[0050] Figures 3a and 3b illustrate that the transmitter above, in a calibration mode, is arranged to transmit a calibration signal.

[0051] Figure 4 illustrates an example of a diagram exhibiting coupling between two sweeps

[0052] Figure 5 is a schematic overview illustrating the principle of received amplitudes mixed with sweep signals in an example of a receiver.

[0053] Figure 6 is block scheme generally presenting an example of a receiver of a communication system

[0054] Figure 7 illustrates schematically transmission and receipt of signals. Figure 8 is an example of a communication system

[0055] Figure 9 is flow chart illustrating an example of a method performed at a transmitter for transmission of a communication signal x.

[0056] Figure 10 is flow chart illustrating an example of a method performed at a receiver for processing of a received signal y transmitted through a communication channel

[0057] DETAILED DESCRIPTION

[0058] The present disclosure will now be described with reference to the accompanying drawings, in which preferred example embodiments of the disclosure are shown. The disclosure may, however, be embodied in other forms and should not be construed as limited to the herein disclosed embodiments. The disclosed embodiments are provided to fully convey the scope of the disclosure to the skilled person.

[0059] Figure 1 illustrates the principle of transmit signal generation at a transmitter. The transmitter comprising a modulator arranged to modulate a communication signal from superposition of a plurality of linear frequency sweeps Xn; n=1,2,

[0060] Each linear frequency sweep has an assigned amplitude (one of ai, a2,..., an) representing information to be transmitted in the particular linear frequency sweep. In an implementation, the plurality of linear frequency sweeps having an amplitude 1 are modified Xn-> anXnthrough assigning individual amplitudes an; n=l,2,... to the respective linear frequency sweeps.

[0061] Each linear frequency sweep has a predetermined frequency range B common for the frequency sweeps. The frequency range is for example 1 MHz or 5 MHz or 25 MHz. The frequency range is for example selected from the range 1-25 MHz, or from 5-25 MHz.

[0062] Each frequency sweep has a predetermined sweep time T common for the frequency sweeps. The sweep time T is larger than the inverse of the frequency range B. With T»l / B, the bandwidth of the linear frequency sweeps Xncan be approximated with B. The time difference between start of consecutive sweeps is the inverse of the frequency range B.

[0063] The respective frequency sweep is carried out during a time interval tn-T / 2<t< tn+T / 2 where tnis denoted as tn=n / B. Thus, the frequency sweeps are displaced in time with a time difference corresponding to the inverse of the frequency range B, i.e. the displacement in time between different frequency sweeps is given as 1 / B. The mathematical consequence of the displacement in time between different frequency sweeps being the inverse of the frequency range, 1 / B, is that the frequency sweeps are close to be orthogonal in relation to each other. This means that the superposed communication signal x has an information bandwidth close to B. The degree of orthogonality increases with the predetermined sweep time T times the frequency range, TB, and becomes perfect when TB -> °°. However in the actual cases of finite TB reconstruction of received amplitudes has to undergo a procedure of signal processing. In his process the transmitted data has to be set-up slightly redundant implying a slight reduction of data transfer rate from B.

[0064] The sweep time is therefore generally selected to be long in relation to the inverse of the bandwidth B. Hence, the sweep time T larger than the inverse of the frequency range B. i.e. preferably, T»l / B. For example, T is in the range 10- 250 times 1 / B. As a consequence of the near orthogonality, the processing on the receiver side will be facilitated significantly.

[0065] A sweep gradient for each frequency sweep is equal to the predetermined frequency range, B, divided with the predetermined sweep time T

[0066] The bandwidth is close to constant with time, the aggregated signal has always the same bandwidth.

[0067] The transmitted communication signal x is formed by the amplitudes of the superpositioned signal at time points tm. This relationship can be written as

[0068] Xm=S{a(tn) Xn(tm); | tm" tn | < T}

[0069] Using this modulation technology, T*B frequency sweeps are transmitted simultaneously.

[0070] In practice, each sweep is modified as Xn-> anXnby assigning the amplitude representing information carried by the respective frequency sweep. Before transmission, the signal is up converted to carrier frequency. The carrier frequency may be 1-10 GHz, for example 5 GHz.

[0071] Before transmission the transmit signal may also be subjected to power amplification and / or band pass filtering.

[0072] Figure 2 is a schematic illustration of an example where N transmit amplitudes (an, n=l, 2.... N) representing information to be transmitted are partly cyclic.

[0073] With the partially cyclic transmit amplitudes, and due to the near orthogonal characteristics, a reconstruction part on the receiver part will be able to perform reconstruction calculations in the frequency domain preferably using a Discrete Fourier Transform, DFT.

[0074] In the illustrated example,, the N transmit amplitudes (an, n=1, 2.... N) comprises a first predetermined number ANi of amplitudes in the beginning of the batch of data and a second predetermined number AN2 of amplitudes in the end of the batch designed to make the batch of data partly cyclic.

[0075] For example, the first predetermined number of amplitudes ANi may equal the second predetermined number of amplitudes AN2, so AN= ANi= AN2 Further, the predetermined number AN of repeated amplitudes may be at least the predetermined sweep time T times the frequency range B, i.e. AN can be written as AN> TB.

[0076] This example is preferably used when the communication channel is without multipath propagation.

[0077] AN corresponds to number of cyclic elements added to handle the coupling between the sweep signals, i.e. to handle the fact that the received signals are not completely orthogonal. As stated above, the degree of orthogonality increases with the predetermined sweep time T times the frequency range, TB, and becomes perfect when TB ->

[0078] In accordance with an example, the batch of data comprising N transmit amplitudes, an, n=1, 2,..., N, is then formed according to below:

[0079] a“aN-3AN+l ' - 'aN-2AN 'al 'aN-3AN+laN-2AN >al / - / aAN

[0080]

[0081] AN element N-2AN element ANelement Thus, the N-2AN transmit amplitudes in the middle section form the information to be carried. The AN transmit amplitudes in the first section is formed from the last transmit amplitudes the in the middle section and the AN transmit amplitudes in the last section is formed from the first transmit amplitudes of the in the middle section.

[0082] Accordingly, it is the transmit amplitudes carrying the information to be carried which is also used for forming the cyclic behaviour.

[0083] In an example, the communication channel has multipath propagation, and the multipath propagation is present with a maximum time spread AT.

[0084] The batch of data comprising N transmit amplitudes, an, n=1, 2,..., N, can then be formed according to below:

[0085] a“aN-2AN2- ANX+1 ' - 'aN- ANX- AN2'al ' 'aN-2AN2- ANX+1 ' - 'aN- ANX- AN2'al ' - 'aAN2

[0086]

[0087] AN-^ elements N-AN-L -AN2elements AN2elements

[0088] wherein ANi corresponds to number of cyclic elements added to handle the coupling between the sweep signals, as described above, and

[0089] wherein AN2 corresponds to number of cyclic elements added to handle the coupling plus multipath propagation.

[0090] The N-ANi- AN2 transmit amplitudes in the middle section form the information to be carried. The ANi transmit amplitudes in the first section is formed from the last transmit amplitudes the in the middle section and the AN2 transmit amplitudes in the last section is formed from the first transmit amplitudes of the in the middle section.

[0091] Figures 3a and 3b illustrate that the transmitter, in a calibration mode, is arranged to transmit a calibration signal.

[0092] The calibration signal may comprise a linear frequency sweep having a known amplitude, the predetermined frequency range B and the predetermined sweep time T.

[0093] Alternatively, or in addition thereto, the calibration mode may comprise buffer zones with no transmission arranged before and after transmission of the calibration signal, wherein the buffer zone before the calibration signal isolates the calibration signal from previous transmissions, said buffer zone having a duration given as (T+AT ), wherein AT represents a multipath propagation spread, if any, and

[0094] wherein the buffer zone after the calibration signal is included to provide the partly cyclic behaviour.

[0095] The shortest possible transmission sequence for calibration consists of a single isolated sweep signal of known amplitude (say =1). On the receiver side, however, such calibration through dispersion and multipath propagation couples to frequency sweeps at times tnthat lie a time T before to a time T+AT after the calibration sweep.

[0096] The calibration sequence therefore is preferably be preceded by a (T+AT) duration 0-amplitude buffer to isolate for data from the previous communication transmission. Further, the cyclicity requirement has the effect that a calibration sequence should then then end with a similar 0-amplitude data buffer, whereby the calibration sequence has at least a duration of 2T+2AT+1.

[0097] Further, since multipath scattering AT is small compared to sweep time T, (2T+2AT+1)B =4TB=4 N applies. Thus, which should normally be allocated for the calibration sequence is approximately 4AN samples long.

[0098] As discussed above, the partial cyclic behaviour also means that the partly cyclic N transmit amplitudes (an, n=l, 2.... N) representing information to be transmitted in a communication batch between two calibrations, consisting of Nc amplitudes there are only NC-2AN independent data amplitudes an. Thus, the transmission efficiency (information bandwidth / signal bandwidth) including the time required for calibration can be written as

[0099] Nr-2AN Nr

[0100] Y = — - - - NcNc+ 2AN

[0101] If disregarding the effect from Doppler, using the transmission above with the discussed length of the calibration period and the discussed added cyclic parts, it is possible to perfectly reconstruct the transmitted signal at the receiver. Figure 4 illustrates an example of a diagram exhibiting coupling between two sweeps with time-bandwidth product T*B=20

[0102] Below please find a description of this coupling in view of Doppler. As stated above, if disregarding the effect from Doppler, using the transmission discussed above with the discussed length of the calibration period and the discussed added cyclic parts, it is possible to perfectly reconstruct the transmitted signal at the receiver.

[0103] However, due to Doppler, errors will always occur.

[0104] Therefore, the number of elements in the Discrete Fourier Transform should be sufficiently small to keep registration time for the data to be Fourier transformed sufficiently short to hold Doppler errors below a given value s.

[0105] The number of elements in the Discrete Fourier Transform may be given as

[0106] Nc= ε / 2π • B / Fc• c / v

[0107] and v is a characteristic value for rate of change of multipath length.

[0108] Thus, if again looking at the transmission efficiency (information bandwidth / signal bandwidth) including the time required for calibration,

[0109] Nr-2AN Nr

[0110] Y = — - - - NcNc+ 2AN

[0111] It is understood that when Ncneeds to be limited due to the effect from Doppler, the transmission efficiency can be lower than desired.

[0112] However, in this situation, wherein due to Doppler, there inherently is an error with a value s, errors in the reconstruction due to coupling between the received signals and multipath propagation may also be allowed. This would increase the transmission efficiency.

[0113] Accordingly, when a finite convolution kernel zmdescribing the communication channel is time dependent, the number Ncof amplitudes to be transmitted between calibrations is limited by an acceptable error E, the buffer zones for a calibration mode can be decreased thereby introducing errors due to the shortening in the same order as errors caused by Doppler.

[0114] In an example, the errors introduced due to shortening is adapted to correspond to noise energy caused by Doppler errors E. In detail, when the wave propagation environment is time-varying, the convolution kernel ■ i2Tn—FC —Vn

[0115] changes overtime whereby received samples bnchange to bn=eB cbn, wherein n / B istime after calibration and v characteristic velocity of change in wave propagation distance. Then, the phase error approximates the relative amplitude error and can be written as

[0116] Frv

[0117] e = 2n - n

[0118] B c

[0119] Typically, an errorB< o.i is required.

[0120] Accepted number of frequency sweeps, and accordingly transmitted amplitudes between calibrations become

[0121] 4n v

[0122] 1 - F

[0123] £CT

[0124] CS

[0125] V= 4n v

[0126] 1+ FC T

[0127] TS

[0128]

[0129] £ c

[0130] The formula generally gives poor transfer efficiency. The reason is the extensive buffer zones required to fully encapsulate multipath propagation and dispersion. Dispersion in this particular context is the delay during which coupling between adjacent sweeps still prevails due to residual lack of perfect orthogonality Such extensive buffer zones are required if an absolutely error-free transmission amplitude reconstruction is sought, which is only relevant in an ideal case without Doppler at all. However, since errors due to Doppler inevitably occur in practice, reducing the buffer zones can be done without performance degradation as long as the accepted errors do not exceed the Doppler errors. As a consequence, transmission efficiency can be dramatically improved, especially at higher Doppler rates.

[0131] When reducing the buffer zones, note that occasional cases of very large delay due to multipath propagation may occur but their contribution to the reception amplitude is negligible because they are so few. It is the energy in the delays that are greater than a certain value AT that gives a certain error s. A typical case is, for example, that in a certain wave propagation environment, signals that through multipath propagation arrive 1 ps after the first signals arrived are rarely present and therefore only contribute - say -20 dB of the received signal energy. Thus, if these signals are excluded when reconstructing transmitted amplitudes, the resulting reconstruction error is -20 dB, i.e. =0.1. The multipath limit to keep the error within s is denoted AE. In the example, is for example AT0 A= Ips.

[0132] Correspondingly, the dispersion can be truncated to a smaller value Tathan T. Let In line with the discussion above relating to Aigthe notation Ta(e) mean that dispersion outside Ta(e) contains a residual energy s2, i.e. that truncation of dispersion to Ta(e) gives a reconstruction error with noise energy s2. In line with the reasoning above relating to that amplitude contributions from large but sporadic multipath delays become small, the energy is small at the edges of the dispersion spread i.e. Ta(0.1) can be expected to be significantly less than T. It can therefore occur that Ta(o.l) < T0 1despite T>> T, making multipath propagation the dominant source of error. Conversely, the dispersion becomes the dominant source of error if Ta(0.1)> AT0.1.

[0133] Summarizing both cases, truncation a time equal to max[AT0 lzTa(0.1)] gives a reconstruction error of agreed order, e «o. l. With dispersion truncated in this way, the transmission efficiency above takes the modified form

[0134] £ R c

[0135] Nc= -, ANX= BTa(0.1), AN2= Bmax AT0 1;Ta(0.1)1 =>

[0136] 2n Fcv

[0137] _ NC-AN1-AN2NC_ l-YfFe {T3(°-l)tmax[AT0il, Ta(0.1)]}

[0138] Nc+ AN, + AN,1 +?5. XF[T(0 1)+ maxriT T(oj)"!}

[0139]

[0140] Further, below please find some calculations relating to energy distribution caused by dispersion. If the energy of a single sweep is 1, the distribution of the scalar product is given squared, i.e. it can be written as

[0141] sin r — i |T - T | I T - T,, | + Tc2

[0142] - LTS J >

[0143]

[0144] HB|T - -T'|

[0145] The total energy contained in the signals received from this sweep during the full dispersion time is equal to the emitted energy, i.e. dr = l

[0146]

[0147] TIBT

[0148] as is indeed an exact mathematical identity.

[0149] When the dispersion time is truncated from Tsto Ta, the energy distributed between Tsand Tawill not be able to be returned to the reconstruction of the original sweep. This energy will thus be reflected in the reconstruction in the form of noise with energy

[0150] = 2B

[0151]

[0152] TIBT

[0153] Inversion of the expression determines the function Ta(s). Numerical evaluation of the expression gives for Ta / Tsthe values 0.5, 0.17 and 0.03 for s=0.1 and respective bandwidths B = 1 MHz, 5 MHz and 25 MHz. Typical use of these formulas is that a given signal bandwidth interval for calibration can be calculated as well as resulting transmission capacity.

[0154] Given the time spread AT for multipath propagation, T is typically chosen to be T»AT. For example, T»10AT.

[0155] Figure 5 is a schematic overview illustrating the principle of received amplitudes mixed with sweep signals in an example of a receiver 600, see figure 6.

[0156] Figure 6 illustrates a receiver 600 for receipt of a signal y transmitted through a communication channel.

[0157] The receiver 600 comprises a linear frequency sweep generator 600 arranged to generate consecutive linear frequency sweeps Ym; m=1,2,... having a common amplitude. Each frequency sweep has a predetermined frequency range B common for the frequency sweeps and corresponding to the frequency range of a transmitted signal, and a predetermined sweep time T common for the frequency sweeps and corresponding to the sweep time of the transmitted signal. The sweep time T is larger than the inverse of the frequency range B.

[0158] The receiver 600 further comprises a data buffer 602 arranged to simultaneously store at least T*B sweeps and to thereby store the generated linear frequency sweeps during their duration. The receiver further comprises a demodulator 603 arranged to sample the signal y received at the receiver to provide sample values y(tn), and to modify each sweep in the data buffer through multiplication with the sample values y(tn) to provide a product y(tn)Ym(tn), and wherein a summation is made for each sample time tnto obtain at least T*B demodulated amplitudes bm.

[0159] The consecutive linear frequency sweeps Ymconstitute the complex conjugates of Xn, with the same sweep gradient.

[0160] Each such sweep is stored in the data buffer for its duration in the receiver. At each time tn, the signal y received from the transmitter is sampled, which gives the sample values y(tn). Each sweep in the data buffer is modified by these sample values y(tn) by multiplication Ym(tn) -> y(tn)Ym(tn). As discussed above, as a final demodulation step, summation is performed over each sweep.

[0161] The demodulated amplitudes bmare accordingly given by

[0162] bm= Sn{y(tn)Ym(tn); |tm - tn| < T}.

[0163] The receiver 600 may further comprise a reconstruction part 604 arranged to reconstruct the transmitted amplitudes (a;, a2,..., an) from the demodulated amplitudes bmbased on knowledge of a finite convolution kernel zmdescribing the communication channel including coupling between the transmitted sweep signals. The coupling between the sweep signals is as discussed above a consequence of the fact that the received signals are not completely orthogonal.

[0164] The reconstruction part 604 may be arranged to reconstruct the transmitted amplitudes (a;, a2,..., an) through convolution with the finite convolution kernel zmaccording to the following:

[0165] bm=S{anZm-n;| m-n|< TB}.

[0166] When the received signal is only delayed, i.e. no multipath propagation and no Doppler, in relation to the transmitted signal, bmcan be obtained from amby convolution with a finite convolution kernel (^0 for | m | < TB) zmaccording to bm=S{anZm-n; | m-n | < TB}. Since the transmit sweeps are approximately orthogonal, the absence of multipath propagation will mean bm~TBam+N for some number N.

[0167] When the N transmit amplitudes (an, n=l, 2.... N) representing information to be transmitted are partly cyclic, the reconstruction part may be arranged to perform the reconstruction calculations in the frequency domain.

[0168] For example, in the case of multipath propagation, since the transmit sweep is only at the distance 1 / B from each other and typical multipath propagation has AT > 1 / B there is in this case not even approximately simple relation between bmand am. In the frequency domain, on the other hand, a simple relation between bmand amexists and can be derived without approximation.

[0169] When the reconstruction is performed in the frequency domain this relies on the N transmit amplitudes (an, n=l, 2.... N) representing information to be transmitted being partly cyclic. In a way the situation can be likened with OFDM and the technique of cyclic prefix added in OFDM symbols to be able to restore subcarrier amplitudes in situations of multipath. In OFDM however cyclic prefix is added to the subcarriers which modulates the transmit signal, while here cyclicity is added only batchwise to the data transmitted.

[0170] With the partially cyclic transmit amplitudes, and due to the near orthogonal characteristics, the reconstruction part will be able to perform reconstruction calculations in the frequency domain preferably using a Discrete Fourier Transform, DFT.

[0171] In an example, the N transmit amplitudes (an, n=l, 2.... N) comprises a first predetermined number ANi of amplitudes in the beginning of the batch of data and a second predetermined number N2 of amplitudes in the end of the batch of data designed to make the batch of data partly cyclic.

[0172] For example, the first predetermined number of amplitudes ANi may equal the second predetermined number of amplitudes AN2, so AN= ANi= N2 Further, the predetermined number AN of repeated amplitudes may be at least the predetermined sweep time T times the frequency range B, i.e. AN can be written as AN> TB.

[0173] This example is applicable when the communication channel is without multipath propagation. The batch of data comprising N transmit amplitudes, an, n=l, 2,..., N, may be formed according to below:

[0174] 3 -aN-3AN+l'-'aN-2AN >al >aN-3AN+l / - / aN-2AN / al / -jaAN

[0175]

[0176] AN element N-2AN element ANelement

[0177] In the given example, AN=TB. In this case, the relation bm=S{amzm-n; | m-n | < AN} can be written as a cyclic convolution:

[0178] AN

[0179] *-*k= znamod(k-n, N-2AN)

[0180]

[0181] n=-AN

[0182] Cyclic convolution is diagonalized by a discrete Fourier transform, which gives the relation D. FkT - - 7ZFkWakT■

[0183] Thus, the above illustrates that the reconstruction part is able to perform reconstruction calculations in the frequency domain for example using a Discrete Fourier Transform, DFT.

[0184] When multipath propagation also occurs with a maximum time spread AT, the following relation is applied: bm=S{amzm-n;-ANi<rn-n< AN2} with ANi=TB, AN2=(T+AT) B. Also in the case of multipath propagation, transmitted amplitudes can be accurately reconstructed from the received amplitudes. This by establishing partial cyclicity according to

[0185] a = aN-2AN2- ANX+1 ' - 'aN- ANX- AN2>al > - >aN-2AN2- ANX+1 > - >aN- ANX- AN2>al > - >aAN2

[0186]

[0187] AN-L elements N-AN-^ -AN2elements AN2elements

[0188] Thereby, the relation between transmitted and received amplitudes given as

[0189] bm=S{anZm-n;-ANi<rn-n< AN2}

[0190] can be written as a cyclic convolution

[0191] AN2

[0192] ^k—,znamod(k-n, N-AN1-AN2)

[0193]

[0194] n=-ANj

[0195] In line with the example above, the cyclic convolution is diagonalized by a discrete Fourier transform, which gives the relation b^ = zkTakT. Thus, the above illustrates that the reconstruction part is able to perform reconstruction calculations in the frequency domain for example using a Discrete Fourier Transform, DFT. The reconstruction part may be arranged to

[0196] perform a discrete Fourier transform on the demodulated amplitudes bm(B[k] = DFT (b[n])),

[0197] provide the convolution kernel in the frequency domain (Z[k] = DFT (z[n] )),

[0198] determine A[k] as B[k] / Z[k], and

[0199] obtain the reconstructed transmitted amplitudes (a;, a2,..., an) as the inverse discrete Fourier transform of A[k],

[0200] A calibration processor 605 may be arranged to determine the finite convolution in a calibration mode from known transmitted amplitudes and known received amplitudes as FT _ FT / FT

[0201] zk -bk / ak ■

[0202] With the finite convolution kernel established, communication data amcan be reconstructed from received data bmaccording to the formula aT= b^T / z^T, as discussed above.

[0203] In different embodiments, the number of elements in the Discrete Fourier Transform is small enough to keep registration time for the data to be Fourier transformed sufficiently short to hold Doppler error below a given value s.

[0204] The number of elements in the Discrete Fourier Transform may be given as

[0205] Nc= ε / 2π • B / Fc• c / v

[0206] and v is a characteristic value for rate of change of multipath length.

[0207] Figure 7 illustrates schematically a modulation scheme as presented herein adopting cyclic ending and commencing of sweep sequences. As illustrated, a transmit signal of N elements has its first 2AN elements repeated at the end. The receive window obtains the N-2AN middle sweeps. Not recording the AN outmost transmit sweeps at either end, also the outwardly directed dispersion of the next AN sweeps is avoided within the receive window. However, by the cyclic character of the sweep sequence the sweeps thus left out are precisely obtained at the receive window opposite end.

[0208] Figure 8 is an example of a communication system implementing the solutions as presented herein.

[0209] The hardware constituents and the distribution of proceeding steps across these will be described in more detail.

[0210] Assume an incoming stream of amplitudes to be communicated, these will be picked up by the transmitter in a data buffer 101.

[0211] The data buffer is for example arranged to hold data for a time longer than the calibration period Tc. The data buffer then allows for intertwining time slots required for calibration. Batches of communication data of required length are obtained from the data buffer 101, whereas calibration i.e. realized by a batch of a single known calibration amplitude 102, inserted into the data stream by a switch 103. To each data package is merged cyclic or more simply zero amplitude extensions of length BTas) at the beginning and Bmax_Azs, Ta(0. / )] at the end 104.

[0212] Each amplitude sample anis thereafter multiplied with an Nslong digital representation of a sweep signal based at a time rnof the sample. Proceeding in this manner there will be Nsamplitudes an,...,an+N-1 associated to each sample moment of rn. Transmit baseband signal at time rnoriginating from a Linear Frequency modulator 105 and a time shift 106 is obtained by summation 107. The result is then digital-to analogue converted in a DA converter, DAC, 109. Transmit signal STX(r) is finally obtained by up conversion to carrier frequency, power amplification and band pass filtering in components 110 - 113.

[0213] Receiving the signal y, initial stages are band pass filtering, a low noise amplification down conversion to complex baseband and AD conversion 202-206, yielding receive amplitudes at sample timesn. Data is thereafter buffered into Nslong stacks, each stack holding amplitude sequences, made up of the amplitudes sampled at times

[0214]

[0215] rn-N, Tn. Each sequence is multiplied by the conjugate

[0216]

[0217] of a sweep signal based at sample time Tn, 207-210. It is noted that the data stacks are incessantly modified, adhering an amplitudes 5^, and removing the amplitude at the passing of sample time rn.

[0218]

[0219] Each stack is summed, outputing the amplitude b

[0220]

[0221] n = ^n^n-Ns)^RX(?n-Ns) + - + ^nMsRXM at each time T„.

[0222] Given the amplitudes bn, transmited communication amplitudes ancan be reconstructed by the DFT based techniques, such as FFT. The reconstruction stage commences with buffering to sufficient length to encompass the calibration period Tc. For reconstruction a precise synchronization should be found within the buffer. This synchronization entails selecting on the receive side the amplitude sequence corresponding to a transmited calibration sequence. Finding this correspondence is suitably enabled by adding a few zero amplitude sweeps at the end of communication sequence, signalling that a calibration sequence is imminent. However, zero amplitude sweeps appear at noise level, the noise uncertainty making a precise matching not viable. On the other hand preciseness is not required since any mismatch just correspond to a signal delay, whereas the delay is compensated for in the reconstruction process. Hence synchronization is obtained to the required level of assuring functionality Note that the latitude in synchronization only holds for determining the transmit-receive delay. The delay between calibration and communication sequences will be given by the parameter set up for the communication scheme and should strictly adhered to, thus obtaining approximate synchronization for the calibration sequence while adhering to the given interval between calibration and communication for applying exactly the same truncation to the calibration and communication data.

[0223] Switch 212 is controlled according to the regularity imposed by the communication scheme adhered too. For either communication or calibration a FFT s applied. The former is present to be zero padded to obtain the length of the communication sequence. Both finding the transfer function kernel from calibration amplitudes and reconstruction transmit amplitude from this kernel has been described in the foregoing and is shown as stages 215 and 219,^ Figure 9 illustrates a method 900 performed at a transmitter for transmission of a communication signal x comprising a batch of data comprising N transmit amplitudes, an, n=l, 2, N, representing information to be transmitted, The method comprises

[0224] modulation 901 of the communication signal x by superposition of N linear frequency sweeps, each frequency sweep having

[0225] an assigned transmit amplitude, an, of the batch representing information carried by the particular frequency sweep,

[0226] a predetermined frequency range B common for the N frequency sweeps, and a predetermined sweep time T common for the N frequency sweeps,

[0227] wherein the time difference between start of consecutive sweeps is the inverse of the frequency range B, and

[0228] wherein the sweep time T is larger than the inverse of the frequency range B.

[0229] Figure 10 illustrates a method 1000 performed at a receiver for processing of a received signal / transmitted through a communication channel, said method comprising

[0230] generating 1 consecutive linear frequency sweeps Ym; m=l,2,... having a common amplitude, each frequency sweep having

[0231] a predetermined frequency range B common for the frequency sweeps and corresponding to the frequency range of a transmitted signal, and a predetermined sweep time T common for the frequency sweeps and corresponding to the sweep time of the transmitted signal,

[0232] wherein the sweep time T is larger than the inverse of the frequency range B,

[0233] storing 2 the generated linear frequency sweeps during a time period long enough to simultaneously store at least T*B sweeps,

[0234] sampling 3 the received signal y to provide sample values y(tn), modifying 4 the stored linear frequency sweeps through multiplication with the sample values y(tn) to provide a product y(tn)Ym(tn), and

[0235] forming 5 a sum for each sample time tnto obtain at least T*B demodulated amplitudes bm- The solutions as presented herein are presented in relation to the problem of providing high transfer transmission efficiency. However, the number of elements comprised in the DFT can be reduced with the requirements of partial cyclicity respected. Such arrangement is obviously also possible.

Claims

Claims1. A transmitter arranged to transmit a communication signal x comprising batches of data each batch comprising N transmit amplitudes, an, n=l, 2, N, representing information to be transmitted, said transmitter comprising a modulator arranged to modulate the communication signal x from superposition of N a linear frequency sweeps, each frequency sweep havingan assigned transmit amplitude, an, of the batch representing information carried by the particular frequency sweep,a predetermined frequency range B common for the N frequency sweeps, and a predetermined sweep time T common for the N frequency sweeps,wherein the time difference between start of consecutive sweeps is the inverse of the frequency range B, andwherein the sweep time T is larger than the inverse of the frequency range B.

2. The transmitter according to claim 1, wherein the N transmit amplitudes (an, n=l, 2.... N) representing information to be transmitted are partly cyclic.

3. The transmitter according to claim 2, wherein the N transmit amplitudes (an, n=l, 2.... N) comprises a first predetermined number Ni of amplitudes in the beginning of the batch of data and a second predetermined number N2 of amplitudes in the end of the batch of data designed to make the batch of data partly cyclic.

4. The transmitter according to claim 3, wherein the first predetermined number of amplitudes ANi equals the second predetermined number of amplitudes AN2, soAN= ANi= AN2, andwherein the predetermined number AN of repeated amplitudes is at least AN=TB.

5. The transmitter according to claim 4, wherein the batch of data comprising N transmit amplitudes, an, n=l, 2,..., N, is formed according to below:a~aN-3AN+l'"-'aN-2AN / al'aN-3AN+l'"~'aN-2AN / al'"~ 'aANAN element N-2AN element ANelement6. The transmitter according to claim 3,wherein when multipath propagation is present with a maximum time spread AT,the batch of data comprising N transmit amplitudes, an, n=l, 2,..., N, is formed according to below,a-aN-2AN2-AN1+l'-'aN-AN1-AN2>alaN-2AN2-AN!+laN-AN!-AN2 / al / - / aAN2AN-^ elements N-AN-! -AN2elements AN2elementswherein ANi corresponds to number of cyclic elements added to handle the coupling between the sweep signals,wherein AN2 corresponds to number of cyclic elements added to handle coupling plus multipath propagation.

7. The transmitter according to any of the preceding claims, wherein the transmitter in a calibration mode is arranged to transmit a calibration signal.

8. The transmitter according to claim 7, wherein the calibration signal comprises a linear frequency sweep having a known amplitude, the predetermined frequency range B and the predetermined sweep time T.

9. The transmitter according to claim 7 or 8, wherein the calibration mode comprises buffer zones with no transmission arranged before and after transmission of the calibration signal,wherein the buffer zone before the calibration signal isolates the calibration signal from previous transmissions, said buffer zone having a duration given as (T+AT ), wherein AT represents a multipath propagation spread, if any,wherein the buffer zone after the calibration signal is included to provide the partly cyclic behaviour.

10. A receiver for receipt of a signal y transmitted through a communication channel, said receiver comprisinga linear frequency sweep generator arranged to generate consecutive linear frequency sweeps Ym; m=l,2,... having a common amplitude, each frequency sweep havinga predetermined frequency range B common for the frequency sweeps and corresponding to the frequency range of a transmitted signal, and a predetermined sweep time T common for the frequency sweeps and corresponding to the sweep time of the transmitted signal,wherein the sweep time T is larger than the inverse of the frequency range B,a data buffer arranged to simultaneously store at least T*B sweeps and to thereby store the generated linear frequency sweeps during their duration, anda demodulator arranged to sample the signal y received at the receiver to provide sample values y(tn), and to modify each sweep in the data buffer through multiplication with the sample values y(tn) to provide a product y(tn)Ym(tn), and wherein a summation is made for each sample time tnto obtain at least T*B demodulated amplitudes bm11. The receiver according to claim 10, wherein the demodulated amplitudes bmare provided as bm=n{y(tn)Ym(tn); |tm — tn| < T}.

12. The receiver according to claim 10 or 11, further comprising a reconstruction part arranged to reconstruct the transmitted amplitudes (a;,a2,..., an) from the demodulated amplitudes bmbased on knowledge of a finite convolution kernel zmdescribing the communication channel including the sweep coupling.l ' l13. The receiver according to claim 12, wherein the reconstruction part is arranged to reconstruct the transmitted amplitudes (a;, a2,..., an) through convolution with the finite convolution kernel zmaccording to the following:bm=S{anZm-n;|m-n|< TB},14. The receiver according to any of the claims 12 - 13, wherein when the N transmit amplitudes (an, n=l, 2.... N) representing information to be transmitted are partly cyclic, the reconstruction part is arranged to perform the reconstruction calculations in the frequency domain.

15. The receiver according to claim 14, wherein the reconstruction part is arranged toperform a discrete Fourier transform on the demodulated amplitudes bm(B[k] = DFT (b[n])),provide the convolution kernel in the frequency domain (Z[k] = DFT (z[n] )),determine A[k] as B[k] / Z[k], andobtain the reconstructed transmitted amplitudes (a;, a2,..., an)as the inverse discrete Fourier transform of A[k],16. The receiver according to any of the claims 14 - 15, wherein the number of elements in the Discrete Fourier Transform is small enough to keep registration time for the data to be Fourier transformed sufficiently short to hold Doppler below a given value E.

17. The receiver according to claim 16, wherein the number of elements in the Discrete Fourier Transform is given asNc= ε / 2π • B / Fc• c / vand v is a characteristic value for rate of change of multipath length.

18. The receiver according to any claims 12 - 17, wherein the finite convolution kernel is determined in a calibration mode from known transmitted amplitudes and known received amplitudes as = b^T / a^T.19 The transmitter according to any of the claims 1-9 or the receiver according to any of the claims 11-18, whereinwhen a finite convolution kernel zmdescribing the communication channel is time dependent, the number Ncof amplitudes to be transmitted between calibrations is limited by an acceptable error E, the buffer zones for a calibration mode are decreased thereby introducing errors due to the shortening, in the same order as errors caused by Doppler.

20. The transmitter or receiver according to claim 19, wherein the errors introduced due to shortening is adapted to correspond to noise energy caused by Doppler errors.

21. A method performed at a transmitter for transmission of a communication signal x comprising a batches of data, each batch comprising N transmit amplitudes, an, n=1, 2,..., N, representing information to be transmitted, said method comprisingmodulation of the communication signal x by superposition of N linear frequency sweeps, each frequency sweep havingan assigned transmit amplitude, an, of the batch representing information carried by the particular frequency sweep,a predetermined frequency range B common for the N frequency sweeps, and a predetermined sweep time T common for the N frequency sweeps,wherein the time difference between start of consecutive sweeps is the inverse of the frequency range B, andwherein the sweep time T is larger than the inverse of the frequency range B.

22. A method performed at a receiver for processing of a received signal / transmitted through a communication channel, said method comprisinggenerating consecutive linear frequency sweeps Ym; m=1,2,... having a common amplitude, each frequency sweep havinga predetermined frequency range B common for the frequency sweeps and corresponding to the frequency range of a transmitted signal, and a predetermined sweep time T common for the frequency sweeps and corresponding to the sweep time of the transmitted signal,wherein the sweep time T is larger than the inverse of the frequency range B,storing the generated linear frequency sweeps during a time period long enough to simultaneously store at least T*B sweeps, andsampling the received signal y to provide sample values y(tn),modifying the stored linear frequency sweeps through multiplication with the sample values y(tn) to provide a product y(tn)Ym(tn), andforming a sum for each sample time tnto obtain at least T*B demodulated amplitudes bm-

Images