GB2431832A

GB2431832A - Multiuser transmitted-reference UWB communications system

Info

Publication number: GB2431832A
Application number: GB0521837A
Authority: GB
Inventors: Klaus Witrisal
Original assignee: Technische Universitaet Graz
Current assignee: Technische Universitaet Graz
Priority date: 2005-10-27
Filing date: 2005-10-27
Publication date: 2007-05-02
Also published as: GB0521837D0

Abstract

An impulse radio (IR) ultra-wideband (UWB) communication system supports multiple users by combining transmitted reference (TR) data coding with three spread spectrum codes: an amplitude spreading code (12, fig. 1), a delay hopping (DH) code (13, fig. 1) and time hopping (TH) code applied by a timing generator (14, fig. 1). The spreading code (33) and delay hopping code (32) reduces narrowband interference, and pulse bursts (34) and idle times (35) are time hopped (36) to avoid collisions (37) with pulse bursts from other transmitters. The receiver (fig. 2) then "tunes" to the desired signal by synchronising (25a, fig. 4) to the three codes by correlation of the received signal with the known code sequences, and applies Forward Error Correction (FEC).

Description

Multiuser Transmitted-Reference UWB Communica-tions System This

invention relates to a wireless ultra-wideband (IJWB) data communications method and apparatus. More specifically it relates to the transmission, reception, detection, and synchronization of a differentially modulated UWB pulse stream, supporting random multiple access and suppressing narrowband interference.

Techniques for UWB communications encode data in the form of signals with cx-tremnely large bandwidths. A con nionly accepted definition of a UWB signal is a signal of an absolute bandwidth greater than 500 MHz or with a fractional bandwidth, defined as the ratio of the handwith and the center frequency, above 20 % [1].

In general, data bearing UWB signals can be generated in many different ways. In this invention, UWB communication systems employing very short electromagnetic pulses are considered, so-called "impulse- radio (IR)" UWB systems. Data can be applied to these pulses for instance by changing their polarity (pulse-amplitude modulation, PAM) or by modulating the pulse timing (pulse-position modulation, PPM).

UWB communication systems have a number of unique advantages over conventional wireless communication systems. Potentially, very high data rates are enabled clue to the large signal bandwidths. At the same time, the transmitted power per unit frequency can be kept very low, which allows the re-use of frequencies assigned to conventional, carrier-based communication systems without interfering with them. That is, it has been suggested to re-use a large part of radio spectrum to employ UWB communications systems at power densities below the emission limits set for un-intended electromagnetic radiation, which applies for instance for switched power supplies, household appliencies, etc. In the United States, the Federal Comrnnnications Commission (FCC) has permitted the application of the frequency band between 3.1 and 11.6 GHz at a power density below -41.3 dBm/MI-Iz for UWB devices [1].

At the same tinie, a communications system at an ultrawide bandwidth can resolve multipath propagation. This implies that the fading margin, required in narrowband wireless systems to cope with multipath fading, can he largely reduced or eliminated from the link budget.

The fine time-resolution enables multiple access assuming that a receiver can he syn-chronized in time to the transmitted pulses at sufficiently lugh accuracy. This is to some extent dual to a carrierbased receiver tuning to the carrier frequency of the desired signal.

On the downside, to exploit these properties of UWB JR signals, receivers for such signals can l)e very complex. A coherent receiver for UWB impulse-radio signals is a rake receiver that needs to generate a local template signal for correlation demodulation of the data. To fully exploit the received signal energy, this template signal must consist of the rnultipath components of the received pulses with a time-accuracy way below the ultra-short duration of the communication pulses. For instance, considering an JR. signal with a bandwidth of 5 GHz, the require(I accuracy is at a magnitude order better than ps, which is extremely hard to achieve and requires fast and power consuming timing generators.

Transmitted reference signaling is au alternative way of encoding data on impulse-radio (or other) UWB signals, yiel(hing sub-optimum, yet simple, receivers. Data are applied by inverting or not the sign of one of two pulses that are transmitted with a time delay D. An autocorrelation receiver frontend is used to demodulate such signals, comprising a delay of (luration D (matched to the delay of the two pulses mentioned before), a multiplier to mnultij.)ly the delayed and not- delayed signals, and an integrator to acci.imulate the signal energy contained in the received pulses. Jn this way, energy from all multipath components is collected without performing any complex chaumel estimation and also the synchronization only needs to be done to an accuracy-level in the magnitude order of the symbol period, which are rather hundreds of nanoseconds in stead of picosecomids [2, 3].

Multiple access can be accomplished by exploiting the selectivity of the autocorre-lation receiver to the time-delay implemented by its delay device. That is, while all energy of the received pulse is collected if the receiver's delay is correctly matched to the delay between the two transmitted pulses, only a rumor, spurious output will result at unmatched delays. By encoding each data symbol on a set of pulses, where consecutive l)ulses are separated by specific sets of delays comprising data chips, and data chips are amplitude modulated by specific spreadling codes, an autocorrelation receiver can accom- 1.)hsh code division multiple access (CDMA) multiuser communication by exploiting its prior knowledge of the specific delay-hopping code and spreading code of the desired user's signal. Such a multiple-access scheme has been termed delay-hopping multiple access communications [2].

Among the disadvantages of autocorrelation receivers are their sensitivity to all kinds of interfering signals within the signal bandwidth, including noise and narrowbarid sig- nials from conventional communications systems. Additionally, although the system ar- chitecture is conceptually very simple, it requires delay elements with very large delay- bandwidth-products and with very well specified delay times. Such delays are very diffi-cult to implement.

The goal of thus invention is the modlificatiomi of a delay-hopping transmitted-reference communications technique such that (i) a higher number of users can be supported with-out catastrophically influencing each other, (ii) the receiver becomes less susceptible to narrowband interference, arid (iii) the autocorrelation receiver front- end can apply shorter delay lines, easing the implementation problem.

The problenis are solved by encoding data in (relatively short) bursts of two or more pulses, which are time-hopped to avoid collisions between signals transmitted by sev-eral users. The pulses within a burst are delay-hopped and modulated by an amplitude spreading code to enable multiple access interference and narrowband interference sup-pnsion.

Essentially, a transmitted signal at a low duty cycle can be generated by encoding data symbols iii short bursts of pulses. That is, the transmitted pulse-burst carrying the symbol energy is followed by a relatively long idle time period, during which the receiver front-end can be disabled. rllhereby the receiver becomes completely insensitive to any interference during this--potentially long-idle time.

On the other hand, during the pulse bursts, the receiver remains susceptible to inter-thring signals. Several strategies are proposed to cope with these interferences.

Error correction coding will correct for erroneously demodulated data symbols caused by occasional, unavoidable collisions between the pulse-bursts of several users, since a global coordination of the (pulse-burst) time-hopping patterns of all users is difficult to achieve and certainly not desirable. Error correction coding is generally implemented in digital communication systems to be able to reliably communicate at lower signal-to-noise power ratios, which extends the effective range of communication systems amid enables the extremmiely low error rates usually required in digital communications.

Narrowband intcrfcrcncc (NBI,) signals will be seen by an autocorrelation receiver as samples of the autocorrelation function of the carrier signal with a fluctuating level due to the mno(lulated signal envelope. These samples follow a cosine-signal at the NBI signal frequency, which is sampled at the delay-lags of the autocorrelation receivers' different delay times. The level of these autocorrelation samples fluctuates according to the signal envelope during the integration interval of the autocorrelation devices. If a pulse is made sufficiently short such that the level of all samples (for different delay lags) is (approximately) equai, then the amplitude spreading code can be designed to average over the sampled cosine function and thereby reduce the impact of the NBI. Post-processing the samples using a digital signal processor, an even more efficient mitigation of NBI is potentially enabled.

A side effect of transmitting the data symbols in relatively short bursts is a reduction of the delog times required in the autocorrelation receiver front-end, simply because pulses need to be transmitted more closely spaced. This simplifies the implementation of the delay elements.

Fig. 1 gives an overall block diagram of the transmitter for the described UWB trans-mission system, comprising several steps of data encoding, UWB pulse generator, pulse timing generator, and transmit antenna.

Fig. 2 shows the receiver comprising the receiver antenna, receiver front-end ampli-fiers and filters, a bank of autocorrelation receiver branches, a back-end signal processor including synchronization processing, a pulse tinling generator, and error correction (he-coding.

Fig. 3 illustrates the encoding of data symbols into time-hopped UWB pulse bursts.

Fig. 3(a) shows a single pulse-burst with delay-hopping and amplitude spreading codes on a pulse level. Fig. 3(b) illustrates the time-hopping code applied to these pulse-bursts, for the desired user (upper trace) and an interfering user (lower trace).

Fig. 4 shows the receiver front-end consisting of a bank of autocorrelation receiver channels, each channel employing a difierent hut fixed delay element.

Fig. 5 illustrates the operation of a synchronization technique for finding the time-hopping code assigned to the pulse bursts.

Fig. 6 illustrates the transmitted burst-oriented UWB signal and an NBI signal having a fluctuating envelope.

Fig. 7 illustrates the response of the autocorrelation receiver front-end to an NBI sigmial when the burst-duration is shorter than the coherence-time of the fluctuating NBI signal.

Fig. 8 illustrates the response of the autocorrelation receiver front-end to an NBI signal when the burst-duration is longer than the coherence-time of the fluctuating NBI signal.

The structure of a typical transmitter for the described UWB commumcations system is illustrated in Fig. 1. Source data 10 is first encoded using any state-of-the-art error correction encoder U output stre in of coded bits (d[j/Nj E {-1, +1}) is then spread by a spreading circuit 12 using a spreading code b E {-1, +i} to express each coded hit by a number of N chips. (Time index j is a time index for the encoded chip sequence. z = [j/Nj is an index for the sequence of encoded data bits, considering that each data hit is expressed by N chips.) Typically, different users employ different spreading codes, which is a well-known means of distinguishing the signals of various users in code division multiple access (CDMA) systems. In this invention, the spreading will be also employed to provide enhanced robustness against narrowband interference, which will be described below.

The next encoding step 13, applied to the spread bit stream, differentially encodes the chip sequence by nniltiplying each chip with the previous encoded chip. Mathematically, the spreading and differential encoding can he writtcn as a1 = a1_ dLj/Nj b1. This sequence (L1 is used in the UWB pulse generator 15 to modulate the polarity of the transmitted UWB pulses 31 according to the data c4 and spreading code b1 33, comprising the UWB pulse stream 30 (see Fig. 3). Typically, N + 1 pulses comprise a pulse-burst carrying one data symbol.

The multipliers 12a and 13a perform binary multiplications, which can alternatively he realized as exclusive-or gates.

A pulse timing gcnerator 14 controls the transnnssion trnies of the UWB pulses, applying the dclay-hopping code and the time-hopping code, which are illustrated in Fig. 3. The delay hopping code sequence {D1} 32 defines the delays between consecutive pulses within a hurst of UWB pulses denoted by 34. Consecutive bursts of UWB pulses are seperated by idle times 35, which are typically longer than the durations of the bursts.

rillies tart times of consecutive bursts of UWB pulses are defined by the time hopping code 36.

Finally, the so-encoded stream of UWB pulses 30, which is generated by the UWB pulse generator 15, is radiated from the transmit antenna.ifi.

A top-level diagram of the receiver is given in Fig. 2. Typically, the UWB pulse stream is captured by the receiver antenna 21, amplified by a low-noise amplifier, filtered by the front-end filter 23 to liniit out-of-band noise and interference, and then applied to a bank of autocorrelation receiver branches 24. These autocorrelation receiver branches are illustrated in detail in Fig. 4. Each branch consists of a delay 41, matched to a delay from the delay hopping code 32, an analog multiplier 42, arid an integrate-and- (lump (I&D) circuit 4. The delay and multiplier undo the differential modulation of a pair of UWB pulses. The energy of this pulse-pair is then collected by the I&D circuit, which is ideally triggered at the arrival time of the second pulse by a timing generator 25, providing one timing pulse per burst, amid a trigger delay 4, triggering the T&D circuits of each of the autocorrelation receiver branches at the second pulses' arrival times. Most efficiently, energy will he collected from a pair of UWB pulses, when its time-separation corresponds to the delay of the autocorrelation receiver branch. If this is not the case, only a small amount of inter-pulse-interference energy will be collected. Therefore, an AcR branch will (approximately) output a value proportional to the spreaded data bit d[J/Njb1 of the pulse-pair separated hy the delay D1 to which the AcR's delay has been matched. That is, the plurality of AcR channels gives at its outputs the encoded chips of a spreaded data symbol. These chips can be efficiently combined by de-spreading the amplitude code by multiplying with the code sequence using a multiplier 2Q and an adder 2Gb. Note that the dc-spreading can he done by a digital signal processor 2Q at the data symbol rate, when the output signals of the AcR channels are converted from the analog into the digital domain.

It is noted that the above described steps of demodulating a delay-hopped, spread, and differentially modulated pulse stream correspond to the delay hopping multiple access scheme described iii the prior art [2]. This method's drawback in the presence of multiple users' signals at the receiver input is a potentially large feedtrough of the signal of an interfering user, which might severely deteriorate the receiver performance. A problem that will he increasingly dramatic considering the near-far-effect, a well-known problem of CDMA systems, and also when the signal bandwidth is not too high, because then time feedtrouglm of interfering users' signals increases. Introducing the time-hopping of commseci.mtive UWB bursts, higher robustness can he achieved even in the adverse scenarios mentioned above. It can be assumed that most symbols (= pulse-bursts) of an interfering user 34a arrive at time receiver during the idle periods 35 of the desired user's signal.

Occasionally however, collisions 37 between the pulse bursts of several users will occur.

In these cases, reliable demodulation of the respective symbols might become impossible.

Nevertheless, the forward error correction decoder 27 potentially enables to correct the errorneously received symbols, yielding the decoded data stream 2().

For a successful recovery of the transmitted data, it is necassary to synchronize time timing generator 25 at the receiver to the time-hopping code fi of the transmitted signal, which is generated by the transmnitter's timing generator 14. This symmchronization is outlimmecl next. The demodulation of the data encoded in a burst of pulses requires knowledge of time delay-hopping sequence {D} and the spreading code {b}, which will be used to "tune" tIme receiver to a certain user's signal by setting the delays 41 and 44 and the de-spreading multipliers 26a accordingly. A such-tuned receiver front-end efficiently collects the energy of each received burst of UWB pulses fromnm tIme desired user, while it suppresses pulse-bursts of interfering users (and narrowband interference). A synchronization algorithm, to be implemented in a synchronization device 25a, can be based omi this capability. In an unsynchronized state, the despread outputs of the receiver front-end c are sampled in evenly spaced intervals. The energy of these samples is deternmmined using a squaring (levice and stored iii somne vector during one full period of the time hopping code, yielding a signal as depicted in Fig. 5(b). This vector is cyclically convolved witim a seleetionm vector, shown in Fig. 5(a), having ones at positions corresponding to the expected arrival times of pulse bursts arid zeros elsewimere. The cyclic commvolutiomm will yield a narrow peak when the phase of the time-hopping code vector matches with the energy values captured, which is evident at sample index 100 in Fig. 5(c). This means, the known time-hopping code to he found (see Fig. 5(a)) starts at sample 100 in the output sample vector (see Fig. 5(b)).

The working principle of the riarrowband interference suppression is illustrated in Figs. 6-8. Fig. 6(a) illustrates a hurst-oriented impulse radio UWB signal. Fig. 6(h) shows a narrowband interference (NBI) signal with a fluctuating envelope. Note that the duration of each pulse-burst is shorter than the typical period of the envelope variations of the NBI signal.

In Fig. 7, the effect of an NBI signal on the output samples of individual samples of the AcR front-end is depicted. 71 is tire autocorrelation function of the NBI signal shown in Fig. 6(h). The dot-markers 7j emphasize the values of this function at the correlation lag times of the individual channels of the AcR front-end. The dashed lines 72 represent the output samples of the AcR channels before applying the despreadimig operations 26a and 26b. It is clearly seen that these output samples in presence of a strong NBI signal follow tire samples 71a of the NBI autocorrelation function. However, their aniplitude variate.s for several UWB pulse bursts, according to the envelope variations of the NBI signal. The de-spreadmg operation coherently combines tire energy of the data components in the output samples, while the energy of the NBI component is spread out and thereby suppressed. It is a task for a proper code-design to devise spreading codes which efficiently suppress NBI. Digital sigmial processing algorithms can be found to optimally collect the energy of a desired signal from such data vectors, while suppressing an interfering signal.

It is a neccessity that the burst durations be shorter than the duration of the enve-lope fluctuations fo tue NBI signal-its cohereirce time-, for a good operation of such NBI mitigation algorithms and for code-based NBI suppression. Fig. 7 illustrates the output sample vectors 82 of an AcR front-end if this condition is not met. Clearly, these samples have a totally random nature, since the samples are now caj.) tured over a time-period longer than the amplitude fluctuations of the NBI signal. Even after despreacling, these noise-like interferences will riot he efficiently suppressed. Neither will simple signal processing algorithms be capable of efficiently mitigating such an interfering signal.

References [1] "Revision of part 15 of the comnnnssion's rules regarding ultra-wicleband transrmussion systems," First Report and Order, ET Doc. 98-153, FCC 02-48, Adopted: February 14, 2002, Released: April 22, 2002, Federal Communications Comnmnission (FCC).

[2] R. T. Hoctor, H. W. Tomlinson Jr., K. B. Welles, and J. E. Hershey, "IJitra-wideband communications system," U.S. Patent 6810087, Jan. 3, 2001.

[3] J. L. Richards, "Apparatus and method for increasing received signal-to-noise ratio in a transmit reference ultra-wideband system," U.S. Patent 0 108 133, June 12, 2003.

Claims

Claims 1. An ultra-widehand (UWB) radio communication system

comprising: a transmitter for generating bursts of N + 1 UWB radio pulses wherein data are encoded differentially in the polarity of said N + 1 consecutive pulses, N = 1, arid wherein pa is of consecutive pulses are separated by Al < N different time intervals and wherein said bursts of pulses are time-hopped within the nominal symbol transmnis-sion period and a rtcei'ver comprising a set of Al pulse-pair correlators 4 for collecting the energy of the differentially encoded data contained in pairs of consecutive pulses wherein the delay lags in said Al pulse-pair correlators are nominally matched to said Al different time intervals that are known to both the transmitter arid the receiver, wherein the outputs of said pulse-pair correlators are combined and wherein a timing generator 25 that is synchrornzed to the time-hopping sequence of the transmitted pulse-bursts enables detection of the data.

2. The transmitter and receiver of claim 1, wherein each burst of pulses represents exactly one binary data symbol.

3. The transmitter and receiver of claim 1, wherein each burst of pulses represents L > 1 binary data symbols by differentially encoding said L data symbols in the polarities of at least L + 1 pulses comprising one pulse-burst, however, L will riot 4. transmitter and receiver of claim I, wherein each binary data symbol is repre-sented by more than one pulse-hurst.

5. The transmitter and receiver of claims 1 4, wherein the data symbols are additionally encoded by a binary spreading code that is known to both tIme transmitter and the receiver, and then differentially encoded in the polarities of the pulse-pairs comprising the pulse-bursts and wherein the outputs of the Pulse-Pair correlators 4 are multiplied by said known spreading code to facilitate constructive combination of the signal energy contained iii multiple pulse-pairs.

6. The transmitter and receiver of claims 1-5, wherein the data symbols are encoded using a forward error correction encoding system 11 and wherein the combined outputs of said bank of pulse-pair correlators are api)lied to a forward error correction decoding system 27 iii order to correct data symbols that were not detected correctly clue to any signal impairments.

7. A niethxxl and apparatus to synchrorize the tinung generator of the receiver to the known, periodic time-hopping sequence imposed by the timing generator of the transmitter 14 on the sequence of transmitted pulse-bursts, comprising a squaring device to extract the received signal energy at uniformly-spaced time instants from the combined outputs of the pulse-pair correlators and a memory device to store said measures of the received signal energy during a time interval that typically matches the period of the time-hopping sequence and a correlator for correlating the stored energy measures to cyclic shifts of the known time-hopping sequence, yielding a correlation peak at a shift value corresponding to the current synchronization time offset, which is consecutively used to align the receiver's symbol clock phase.

8. The syncronization method and apparatus from claim 7 wherein the combined outputs of the pulse-pair correlators are directly stored in said memory device and wherein the stored sequence is correlated to cyclic shifts of the known time-hopping sequence incorporating a known data sequence (training sequence), again yielding a correlation peak at a shift value corresponding to the current synchronization time offset, when said training sequence is transmitted.

9. The communication system of claims 1 6, wherein multiple similar transmitters are simultaneously transmitting signals com- prising (typically unsynchronized,) differently time-hopped sequences of UWB pulse-bursts and wherein a receiver is synchronized to the timne-hoppimig sequence of one particular "desired" signal using the synchronization method from claim 7 or claim 8, facilitating the detection of the desired data stream and the discrimination of the other transmitted signals (multiple access).

10. rp multiple access communication system of claim 9, wherein each transmitter encodes the transmitted data on the pulse bursts using different sets of time intervals for separating consecutive pulse-pairs (within a pulse-burst) and/or superimposing diflerent spreading codes on the data that is differentially encoded in the polarities of said pulse-pairs, thereby improving the capability of discriminating the signals of the non-desired users.

11. The multiple access communication system of claim 9 or claim 10, using a digital signal processor 26 in a way that optirrially combines the received signal energy of the desired signal and/or suppresses the received signal energy from the non-desired signals (rnultiuser detection or multiple-access-interference (MAT) suppression).

12. The transmitter and receiver of claims 1-6, wherein the spreading code is designed such that the receiver spreads out (i.e. reduces) the energy of an interfering nar-rowband signal wiule dc-spreading (i.e. collecting) the energy of the desired signal, thereby increasing the signal-to-interference power ratio at the output of the com-biner 26.

13. The transmitter arid receiver of claim 12, using a digital signal processor in a way that optimally combines the received signal energy of the desired signal and/or suppresses the signal corresponding to the narrowband interference.

14. The receivers frommi claims 1-6 using the digital signal processors as described in ii arid 13, wherein the output signals of the hank of pulse-pair correlators 24 are sampled K-times Ier symbol period, allowing for more efficient supression of NBI and MAT.