SYSTEM

TECHNICAL FIELD

The invention relates to theimplementation of an active distributed antenna system (DAS) transporting digital passband data

BACKGROUND

A distributed antennasystem (DAS) is a technology for providing radio coverage in an area which cannot be directly served from a radio base station, and is particularly advantageous in applications where multiple wireless service providers need to provide coverage si nee a single DAS can be used with many radio basestations.

A state of the art DAS 100 uses digital transport for radio signals as shown in Figure 1. For the sake of brevity, only onedirection of thesignal path is shown and for asingleband only. In practice, a DASisbidirectionaJ, carrying downlink signaJsfrom radio basestations (RBS) 101-104 as shown in Figure 1 to co ver age ar eas p r o vi d ed by remote antennas 105, 106 for serving wireless communication devices (not shown), such as smart phones and tablets.

Uplink signals are carried in theoppositedirection, from the wireless communication devices to the RBSs. A DAS will usually support multiple frequency bands in both downlink and uplink.

I n general, the DAS 100 consists of oneor moresourceunits 107, 108, which each interface to oneor more base stations 101-104, and oneor more remote units 109, 11o which drive the antennas 109, 110 in the respective coverage area. I n between thesourceunits 107, 108 and the remote units 109, 110 there may be some type of routing unit 111, either as oneor more separate units, or integrated into thesourceand remoteunits, or some combination thereof. I n the example of Figure 1, a single routing unit 111 is shown. In the example DAS of Figure 1 there is analogue feed, and the base station input signalsfor a frequency band are presented to each sourceunit as a combined signal at a respective analogue-to-digital converter 112, 113 (ADC).

The bandwidth of thesignal into the ADCs 112, 113 isdefined by the minimum and maximum frequencies all owed for base station input signaJsin the band of interest. As an example, the 1800MHz Digital Cellular System (DCS) cellular band hasadownlink frequency range of 1805.2 to 1879.8MHz. Thisistypically amuch wider band than that of any individual signal from the RBSs 101-104, and the ADC sample rate must be high enough to sample theentireinput signal band. As an example, in order to avoid aliasing, the sample rate of theADCs 112, 113 must exceed twicethe total bandwidth, i.e. in this particular example 2 ^* (1879.8-1805.2) = 149.2 MHz.

A number of channel filters 114- 117 separate the individual channels of the respective base station 101-104 into independent streams of samples. These individual streams of samples corresponding to signals transferred over each channel of the RBSs are scheduled and serialized by ascheduler 118, 119 and a serial izer 120, 121 for transmission over high speed digital links 122, 123 such as Common Public Radio I nterface (CPRI ) links using fiber-optic connections. As can be concluded, this is problem aticfor thescheduler 118, 119 and the serial izer 120, 121, sincethe ADCs 112, 113 must sampleincoming data at ahigh sample rate, resulting in a great amount of digital data arriving at thescheduler/ serializer.

At therouting unit 111, thesamplesarede-serialized 124, 125 and passed onto a routing function 126 which forwards the samples to therequired output ports. At each output port, theset of samples destined for that port are again scheduled and serialized by ascheduler 127, 128 and a serializer 129, 130 for transmission over high speed digital links 131, 132.

Finally, at each remoteunit 109, 110, thesamplesarede-serialized 133, 134 and passed to transmit filtering functions 135-138, which areconfigured to recreate the original radio signalsfor each channel being transported. The outputs of all thetransmit filtering functions 135-138 for afrequency band aresummed and passed to a respectivedigital-to-analogue converter 139, 140 (DAC) to recreate an analogue signal which can be amplified and transmitted over theantennas 105, 106 providing the coverage areas serving the wireless communication devices.

Sourceunits 107, 108 can also bedesigned which have a purely digital interface to each base station 101-104, in which case the signals to be transmitted are transmitted in theform of digital samples. In thiscase, the role of the receive filtering 114- 117 i s to format and convert thedigital samples from the base stations 101-104 into aformat suitablefor

transmission through the DAS 100.

Sgnalsin the"real" world exist over a finite range of frequencies, such as radio frequency (RF) signalstransmitted by the respective base station 101- 104, and when sampled (e.g. by theanalogue-to-digital converter 112, 113) can be represented by a sequence of digital samples.

US patent number 8,929,288 discloses a DAS including a host unit and a plurality of remoteunits. The host unit includesaplurality of base

transceiver stations and a switch. Each of the base transceiver stations is configured to provide a downstream baseband digital signal to the switch and to receive an upstream baseband digital signal from theswitch, wherein each downstream baseband digital signal and upstream baseband digital signal is adigitaJ representation of theoriginaJ radio frequency channel at baseband of the respective base transceiver station. Theswitch isconfigured to route each of the downstream baseband digital signalsto a respective subset of the remote units as one or more downstream serial data streams and to route each of the upstream baseband digital signalsfrom oneor more upstream serial data streams to a respective subset of the base transceiver stations. Further, the standardized Common Public Radio I nterface (CPRI ) interface specification "Common Public Radio I nterface (CPRI), I nterface

Specification", currently Version 6.0, advocates serialization of baseband data over a high speed digital link between a baseband unit and a radio head. DigitaJ DASimplementations have followed this approach, such as the system described in US8,929,288 where baseband samples are serialized over a digital link.

Processing the digital data at baseband in the DAS has its advantages. With reference to Figure2, whereadigital baseband representation of the origi nal radio frequency signal is illustrated, it can be seen that with such a digital baseband representation each data sample is represented as a complex number (consisting of a real and an imaginary components). It is noted that thereal and imaginary components of a baseband signal practically extend from 0 Hzto ahigher cut -of frequency, such asjust below 10 kHz. Negative frequencies are mirrors of the corresponding positive frequency components.

The well -known Nyquist ^' s theorem teaches that as long as the occupied bandwidth of a signal islessthan half of thesample rate F _s then the analogue signal can be perfectly reconstructed from the stream of digital samples.

As can be seen in the left-hand power spectral density (PSD) illustration of Figure2, thecombined complex representation allows for a two-sided power spectrum where negative frequency componentscan be different from positive frequency components, allowing a total bandwidth up to thesample rate F _s .

From a signal processing perspective such a representation iscommonly used because, among other reasons, as shown in the right-hand timedomain illustration, it allows the real and imaginary components to be processed in parallel. DigitaJ baseband signals are commonly referred to as I Q data, i.e. user plane information in theform of in-phaseand quadrature modulation data. This IQ modulation enables thedigital data to be represented by the real and imaginary components shown in the right-hand timedomain illustration of Figure2.

A DAS needs to deal with a wide range of signals corresponding to different cellular operators and mobiletransmission standards. This meansthat a wide range of different signal bandwidths may be presented, for examplewithin thetotai ADCinput bandwidth. The minimum sample rate for a wideband signal is larger than that for a narrowband signal, and so the range of bandwidthswhich must be supported leadsto theneed to support a wide range of different sample rates to makeefficient use of the available capacity of thedigitaJ interconnection. This means that thescheduling process located before each serial izer has a challenging task; it is necessary to find a schedule for transmitting datasamples for each channel while adding a minimum amount of delay. At each step through the DAS where it is necessary to carry a different mix of samples (for exam pi eat each intermediate routing step) it is necessary to provide further buffering to compensate for the extra scheduling delay introduced. Thetotai delay can be critical to the

performance of a DAS si nee there are limits to how much the base stations can be adjusted to compensate for thedelay through the DAS.

The flexibility of the schedulers 118, 119 of the DAS 100 in Figure 1 islimited by thegranularity of thedatato be transmitted. A baseband data sample consists of two sample values as was illustrated in Figure2, thereal and the imaginary component, which must both be serialized 120, 121 in order to be transmitted over the digital links 122, 123. This meansthat the minimum granularity for the schedulers 118, 119 consists of thetimetaken to process both components.

A drawback with the di gitai baseband representation discussed in the above is that there isdel ay associated with the processing of each timedomain digital data component. SUMMARY

An object of thepresent invention isto solve, or at least mitigate, this problem in the art and to provide an improved DAS and a method of transporting digital data in the DAS.

This object is attained in afirst aspect of theinvention by a method of transporting digital data i n a DAS. The method comprises receiving data from at least one data source, processing the received data, and providing the processed data as digital real -valued passband data for further transport within the DAS.

This object is attained in a second aspect of theinvention by a device configured to transport digital data i n a DAS, thedevice comprising a processing unit and a memory, said memory containing instructions executable by said processing unit, whereby said device is operative to receive data from at least one data source, process the received data, and providethe processed data as digital real-valued passband data for further transport within the DAS.

As previously has been discussed, when the real and imaginary component of a com pi ex digital signal isscheduled and serialized in a DAS, a processing delay occurs at the processing of the respective signal components.

Henceif thedeiay for processing a single one of thereal and theimaginary signal component isdenoted D, is follows that thedeiay for processing both components of each sample is 2 ^* D. I nevitably, both components must be processed in order to betransmitted over thehigh speed links of the DAS.

This problem is advantageously overcome by a method of transporting digital data i n a DAS usi ng a passband representation according to the i nvention. In contrast to a baseband representation, a passband representation does not consist of independent real and imaginary components, but rather can be represented by a single component, such asapurely real-valued signal. I n the DAS proposed with the embodiments of theinvention described herein transporting a real-valued digital passband signal, the processing delay of the schedulers and serializersfor preparing transport of each data sample over thehigh speed datalink is reduced by 50%, and hence amounts to a processing delay of D for each digital passband data sample as compared to a delay of 2*D as would bethecasefor each digital baseband data sample consisting of a real and an imaginary component.

In an embodiment, theprocessing of the received signal includes filtering, in what isref erred to as an RX filter, each channel provided by the base stations (or in case of uplink communication; provided by wireless communication terminals). Hence, each respective channel of the base stations is processed by a corresponding frequency-selective RX filter. In thiscontext, a channel should be construed asaset of signals occupyi ng a range of frequencies which the operator wishes to transfer together through the DAS.

Thereafter, thesignalsof each filtered channel is re- sampled or decimated in the respective RX filter in order to reduce the sample rate F _s of thesignal, where t he sample rate Fs is adapted to the bandwidth of thesignals transported over the respective channel. As can be concluded, thesamplerate Fs applied in the RX filters must beat least twice the bandwidth of the highest-bandwidth signal of the fi !tered channel.

Thisis highly advantageous, as the resulting sample rate F _s of the real -valued digital passband data provided by each RX filter for further transport within the DAS is adapted to the actual bandwidth of thefiltered channel, rather than thetotal bandwidth of thesignalscoming into theADCs 112, 113.

Thefiltered base station si gn al s h ave a n ar rower bandwidth, and can be represented with a lower sample rate than the ADC sampling the incoming composite signal of the base station. For efficient transfer over the di gitaJ link, thechannel filtering thusalso in an embodiment includes oneor more stages of decimation or resamplingto reduce the sample rate of each base station channel.

Asaresult, by re-sampling/ decimating the signal of the respective filtered base station channel at a lower sample rate F _c , using only as high sample rate as necessary taking into account the bandwidth of the fi itered channel, the amount of digital passband data provided to theschedulers is greatly reduced whilestill enabling subsequent reconstruction of the origi nal signal at the remoteunitsfor transmission over theantennas. Thisgreatly mitigatesthe processing burden on the schedulers/ serial izers.

Further embodiments of theinvention will bediscussed in thedet ailed description.

Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in thetechnical field, unless explicitly defined otherwise herein. All references to "a/ an/ the element, apparatus, component, means, step, etc." areto be interpreted openly as referringto at least one instance of the element, apparatus, component, means, step, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not have to be performed in the exact order disclosed, unless explicitly stated.

BRIEF DESCRIPTION OF THE DRAWINGS

Theinvention is now described, by way of example, with reference to the accompanyi ng drawi ngs, in which:

Figure 1 illustratesa prior art DAS system in which the present invention may be implemented;

Figure2 illustrates digital baseband data in thefrequency domain and time domain;

Figure 3a illustrates a flowchart describing a method according to thebasic idea of theinvention; Figure 3b illustrates aflowch art describing a method according to an embodiment of theinvention;

Figure4 illustrates real -valued digital passband data in the frequency domain and timedomain;

Figure 5 illustrates a flowchart describing a further embodiment of a method according to theinvention;

Figures 6a-d illustrates transforming digital baseband data to real-valued digital passband data for further transport within the DAS according to an embodiment;

Figure 7 illustrates a flowchart describing yet a further embodiment of a method according to theinvention;

Figure8 illustrates aflowch art describing still a further embodiment of a method according to theinvention;

Figure9 illustratesaflowchart describing still another embodiment of a method according to theinvention; and

Figure 10 illustrates a device according to an embodiment of theinvention performing the methods disclosed herein.

DETAILED DESCRIPTION

Theinvention will now be described morefully hereinafter with reference to the accompanying drawings, in which certain embodiments of theinvention are shown. This invention may, however, be em bodied in many different forms and should not be construed aslimited to the embodiments set forth herein; rather, these em bodi ments are provided by way of example so that thisdisclosurewill be thorough and complete, and will fully convey the scope of theinvention to those skilled in theart. Like numbers refer to like elements throughout thedescription. Figure 1 illustrates a prior art DAS in which the present invention may be implemented. The DAS 100 of Figure 1 has previously been discussed in detail.

Figure2 illustrates a frequency domain and atimedomain representation of digital baseband data. X _r (n) denotes the real component of the digital signal in thetimedomain at discrete time sample index n, whileX,(n) denotes the imaginary component of thedigitaJ signal in thetimedomain at discretetime sample index n.

As previously has been discussed, when the respective component of the digital signal isscheduled and serialized by theschedulers 118, 119, and the serializers 120, 121 of thesourceunits 107, 108 and subsequently by the schedulers 127, 128 and serializers 129, 130 of therouting unit 111, a processing delay occurs at the processing of the respective signal

components.

Henceif thedeiay for processing a single one of the two digital signal components is denoted D, isfollowsthat thedeiay for processing both components of each digital sample is 2 ^* D. Inevitably, both components must be processed in order to betransmitted over thehigh speed links 122, 123 and 131, 132.

This problem is advantageously overcome by a method of transporting digital data i n a DAS usi ng a passband representation. I n contrast to a baseband representation, a passband representation does not consist of independent real and imaginary components, but rather can be represented by a single component, such as a purely real-valued signal.

A passband signal occupies a band of frequencies which areonly positive (or, equivalently, only negative). It is not necessary that negative and positive frequency components are i ndependent from one another, and thereforethe signal does not need both real and imaginary components in order to be represented. A real-valued digital passband signal isdefined as being a signal where each sample is represented by a single coordinate. This impliesthat the possible values for samples of thesignal map to alinein the complex plane. This line is typically the real axisin the complex plane, but could equally be the imaginary axisor a straight line at any other position.

The method of the invention comprises, with reference to theflowchart of Figure 3a, receiving data in step s101 from at least one data source, which data source may be embodied by the radio base stations (RBSs) 101-104, but could alternatively bean inter mediate device which receives an RF signal from the base station and mixesthe RF signal down to a lower frequency, such as intermediate frequency (I F), and/ or converts the anal ogue RBS signal to adigital signal.

In thefollowing, it isassumed that any mixingof an RF signal down to a lower-frequency I F signal is performed by an RF mixer/demodulator located between the respective RBS 101-104 and the DAS 100, or within the DAS 100 as illustrated with the mixer/demodulator 141, 142.

The RF signal istypicaJiy a passband anal ogue signal, which when supplied to a DAS (via any mixer device) commonly is a composite of signalsfrom oneor more RBSs 101-104, even though a DAS may receive signals from a single RBS. However, in practice, a DAStypically receives signals from numerous different RBSs, which may be operated by various operators each operating at different frequencies. These different channels must be separated by the respective channel RX filter 114-117, and subsequently reconstructed by the respective channel TX filter 135-138 at the remote units 109, 110.

Thereafter, the received data is processed in step S102 accordingly, depending on which type of data it comprises- analogue, digital, RF, I F, baseband, etc. - and the processed data is provided in step S103 to the respective scheduler 118, 119 as real-valued digital passband data for further transport within the DAS 100. In thedownlink, the processing as illustrated in steps S101-S103 isin an embodiment performed in the source units 107, 108 closeto the RBSs 101- 104 for further downlink transport, while in theuplink theprocessing of steps S101-S103 is performed in theremoteunits 109, 110 for further uplink transport. Hence, when data enters the DAS 100- either in an uplink or downlink direction - thedata is processed such that it can be provided as real-valued digital passband data for further transport within the DAS 100.

With reference to Figure 3b, in an embodiment, theprocessing step S102 of Figure 3a includes the further step S102b ^* of filtering, in each RX filter 114- 117, each channel provided by the RBSs 101-104. Hence, each respective channel of the RBSs 101-104 isprocessed by a corresponding RX filter 114- 117.

Thereafter, in step S102b", thesignalsof each filtered channel isre- sampled or decimated in the respective RX filters 114-117 in order to reducethe sample rate F _s of thesignai, wherethesample rate F _s is adapted to the bandwidth of the signals transported over the respective channel. As can be concluded, the sample rate F _s applied in the RX filters 114-117 must be at least twice the bandwidth of the highest-bandwidth signal of the fi Itered channel.

Thisishighly advantageous, as the resulting sample rate F _s of the real -valued digital passband data provided by each RX filter 114-117 for further transport within the DAS 100 is adapted to the actual bandwidth of thefiltered channel, rather than thetotaJ bandwidth of thesignaiscoming into the ADCs 112, 113.

Thefiltered RBS signals havea narrower bandwidth, and can be represented with a lower sample rate than the ADC 112, 113 sample rate. For efficient transfer over the digital link, thechannel filtering thusalso in an

embodiment includes oneor more stages of decimation or resamplingto reduce the sample rate of each RBS channel. Asaresult, by re- sampling/ decimating the signal of the respectivefiltered RBS channel at a lower sample rate F _s , using only as high sample rate as possible taking into account the bandwidth of thefiltered channel, the amount of digital passband data provided to theschedulers 118, 119 is greatly reduced whilestill enabling subsequent reconstruction of the origi nal signal at theremoteunits109, 110 for transmission over theantennas 105, 106. This greatly mi ti gat est he processing burden on the schedulers serial izers.

As previously has been described, at each remote unit 109, 110, thesamples are de- serialized 133, 134 and passed to transmit f ilteri ng functions 135-138, which areconfigured to regenerate the ori gi nal radio signalsfor each channel being transported, including interpolation of the sample rate F _s of each signal to a high enough sample rate such that the entire output frequency band corresponding to the ADC input frequency band can be recreated. The outputs of all thetransmit filtering functions 135-138 for afrequency band aresummed and passed to a respectivedigital-to-analogue converter 139, 140 (DAC) to recreate an analogue signal which can be amplified and transmitted over theantennas 105, 106 providing the coverage areas serving the wireless communication devices.

Figure4 illustratesa real-valued digital passband signal in thefrequency domain and in thetimedomain, respectively. A digital passband signal occupi es a sped f i c band of positive frequencies in thefrequency domain as shown in the left-hand illustration. The power spectrum at negative frequencies is an identical mirrored version of that at thepositive

frequenci es.

As shown in the right-hand timedomain illustration, the di gi taJ signal can be represented by a sequence of real-valued samples, as shown in the right-hand illustration. Y _r (2n) denotes the real component of the digital signal in the timedomain at discrete time sample index 2n. Again, according to Nyquist's theorem; as long as the sampling rate F _s isat least twicethe bandwidth of thesampled analogue signal, the analogue signal can be perfectly reconstructed from the stream of digital samples. The factor 2 for the sample index indicates that the sample rate for apassband representation must be twice that of a baseband representation with the same information bandwidth.

When comparing the real-valued digital passband signal of Figure4 to the digital baseband signal illustrated in Figure2, it can be concluded that the digital baseband signal representation enables parallel processing of the real and imaginary components of a data sample, with half thesampling rate F _s of a passband signal having the same bandwidth.

However, in the DAS proposed with the embodiments of theinvention described herein transporting a real-valued digital passband signal, the processing delay of thescheduler 118, 119 and serial izer 120, 121 for preparing transport of each data sample over thehigh speed data link 122, 123 isreduced by 50%, and hence amounts to a processing delay of D for each digital passband data sample as compared to a del ay of 2 ^* D as would bethe casefor each digital baseband data sample consisting of a real and an imaginary component.

An embodiment of a method of transporting digital data in a DAS according to theinvention will now be described with reference to the flowchart of Figure 5 and the signals illustrated through Figures 6a-d.

According to the embodiment, real-valued digital passband data is to be transported by the serial izers 120, 121. In this particular embodiment, with reference to the flowchart of Figure5, theinput to thesourceunits 107, 108 of the DAS 100 is an RFsignal which is mixed down to an analogue baseband signal with zero intermediate frequency with a quadrature

mixer/demodulator 141, 142 in step S102a. This quadrature

mixer/ demodulator may be located within or outside the source units 107, 108. Hence, it may be envisaged that the DAS 100 receives an I F signal already having been mixed down and quadrature demodulated by an external quadrature mixer/ demodulator, in which scenario the DAS 100 would not perform step S102a.

I n aquadrature mixer/demodulator, two mixers are utilized which mix the sameinput signal with two different versions of a local oscillator signal which are 90 degrees off set from one another, which produces an output signal which can be treated as a complex representation.

The ADCs 112, 113 will thus sample the analogue I F signal as a complex baseband signal in step S102b.

Oncethe I F signal has been digitized, the processing consists of separating thedifferent signalsthat makeup the composite input signal by frequency- selective filtering as has been discussed hereinabove with reference to Figure 3b in step S102b'. Hence, each channel has a frequency-selective digital filter 114-117 whose bandwidth is adapted to the character! sties of the respective signal source, i.e. each filter istuned to the operating frequency of the corresponding RBS.

As has been previously described, thisadaption of thesignal bandwidth is performed by re- sampling or decimating thesignalsof the fi Itered channels, (step S102b" of Figure 3b), advantageously resulting in a I ower sample rate Fs. It should be noted that this sample rate Fs is substantially lower than that applied by the ADCs 112, 113, which need to accommodated the total bandwidth of signalscoming to the DAS 100 over all RBS channels.

Now, after thisfiltering, a digital baseband signal - as illustrated in Figure 6a (which previously has been discussed with refer en ceto Figure 2) - has been obtained for each channel resulting in a digital baseband signal with sample rateF _s (where F _s may be different for each channel, depending on thesample rate needed to represent the filtered signals without aliasing). This digital baseband signal isin this particular embodiment then

transformed into a real -valued digital passband signal using a3-step approach starting with the step of performing upsampling in step S102c. This step can be omitted if the di gi tal baseband signal already has a sample rate that is at least twice the bandwidth of the information contained in thesignal.

In thisexample, upsampling is performed with a factor 2, i.e. the digital baseband signal is interpolated by a factor 2, thereby doublingthesample rate and increasing the Nyquist bandwidth to 2*F _S .

Effectively, when performing upsampling, zeros are inserted between the original samples to increase the sampling rate, followed by lowpass filtering to smooth theresulting upsampled digital signal, thereby reconstructing the wanted signal across the inserted zeros.

Figure6b showsthedigital baseband signal being the result of the

upsampling of the digital baseband signal of Figure 6a. It should be noted that theupsampling adds undesired spectral images (indicated with dashed lines) to thesampled signal, which spectral images are centered on multiples of the origi nal sampling rate and must be removed by the lowpass filtering mentioned previously. Thus, with theupsampling, it becomes possible to represent signalsin afrequency rangefrom -F _s to +F _S . However, thedesired signal still liesin the origi nal frequency range of -Fs/2 to +Fs/2.

After theupsampling of step S102c, as is illustrated in thetimedomain representation in the right-hand side of Figure6b, thesameamount of data is present in theupsampled digital baseband signal as compared to the origi nal digital baseband signal of Figure 6a, but thesampling rate has increased to

Thereafter, afrequency shift of theupsampled digital baseband signal is performed in step S102d by afrequency Fs/4 (of theupsampled frequency, corresponding to Fs/2 with respect to the origi nal sample rate applied by the ADCs 112, 113), thereby moving the lower edge of thesignal to a positive frequency, in order to create a complex digital passband signal. The result of thisoperation is illustrated in Figure 6c. It isalso possibleto choose a different frequency shift, as long asthe result is a passband signal (i.e. the wanted signal is enti rely located at positive frequencies or entirely at negative frequencies).

Finally, the imaginary component of thecomplex digital passband signal illustrated in Figure 6c isdiscarded in step S102e, resulting in thesignal illustrated with reference to Figure 6d, i.e. the real-valued digital passband signal previously discussed in detail with reference to Figure4 is provided as illustrated with step S103. As compared to the di gitaJ baseband signal of Figure 6a, which iscentered at zero frequency, the digital signal of FigureSd is centered at Fs 2 and consists of a passband of positive frequencies on either side. Without the imaginary component, the negative frequency range becomes a mirror image of the positive frequency range, which means that uniqueinformation only can be carried in thefrequency rangeO to+Fs;

however, it has already been madesurethat thedesired signal liesin that range. Alternatively, thereal component can bediscarded; this hasthesame effect. A center frequency other than Fs 2 may be used as long as there is enough margin in the sample rate such that the enti re signal bandwidth is still located in the positive (or negative) half of thespectrum.

Theresult is a signal with the same average data rate - i.e. half theamount of data, twice as often - but with a scheduling granularity that advantageously is half that of a baseband representation, si nee each of the data samples is real - valued and can be scheduled and routed independently.

Hence, as compared to the digital baseband representation, where one real and one imaginary component must be processed in sequence for each data samplewith the corresponding delay, the processing delay for each data sampleof the real-valued digital passband signal when scheduling and serializing the data samples is reduced by 50%. At the remote units 109, 110, before transmitting the original RF signal received from the RBSs 101-104 via the antennas 105, 106 to any wireless communication devi ces, the inverse of the above described 3-step approach is performed at each of thechannel TX filters 135-138; a baseband signal can be recreated from thepassband signal of Figure 5d by

(1) shifting thefrequency back to baseband, and

(2) f iiteri ng the resulting complex signal to remove the unwanted imagefrom theshifted negative frequency component, and

(3) downsampling (i.e. reducing) the sample rate F _s by a factor 2.

Finally, after having transformed the digital signal of each respective channel back to the baseband representation shown in Figure 5a, thechannel TX filters reconstruct thesignalsof each channel such that they can be summed together with the other channels and converted from digital format to analogue format by the DACs 139, 140 before passing through aquadrature mixer/ modulator 143, 144 at each remote unit 109, 110 after which the RF signals as provided by the RBSs 101-104 are regenerated (or created for the first timein theevent of a digital baseband feed to the DAS 100) and transmitted viatheantennas 105, 106.

In a practical implementation, it ispossibleto integrate some or all of the conversion steps i nto existing signal processi ng ci rcuitry (sincethe receive path typically involves stages of frequency shifts and decimations, with the opposite steps in thetransmit path) if use of baseband data representation is desired for internal processing. Alternatively, an implementation could choose to useapassband representation throughout.

A series of transformations to passband representation then back to baseband representation does not provide an identical baseband signal at the receiving end (si nee the filtering steps involved are non-ideal, and the process is not time-invariant). However, Nyquist's criterion is never violated so the essential properties of thesignal are preserved and it ispossibleto recreate a RF signal with arbitrarily high fidelity at theoutput.

I n an alternative embodiment, described with reference to the flowchart of Figure 7, the input to the source units 107, 108 of the DAS 100 isan RF signal having been mixed down to I F - but not to zero but to any other appropriate intermediate frequency - or being mixed down and quadrature modulated within the DAS 100 with thequadrature mixer/ demodulators 141, 142 in step S102a.

The ADCs 112, 113 will thus sample the analogue I F signal as a complex passband signal in step S102b.

Again, oncethe I F signal has been digitized, the processing consists of separating the different signalsthat makeup the composite input signal by frequency-selective filtering in RX filters 114-117 (step S102b' of Figure3b). Hence, each channel has a frequency-selective digital filter 114-117 whose bandwidth is adapted to the characteristics of the respective signal source, i.e. each filter istuned to the operating frequency of the corresponding RBS.

Further, the signalsof the fi Itered channelsare re- sampled/ decimated at a reduced sample rate Fs as previously discussed in step S102b" of Figure 3b, which sample rate Fs takes into account the actual bandwidth of thesignals of the fi Itered channel.

Now, after thisfiltering, a complex digital signal comprising both real and imaginary components - as illustrated in Figure 6c- has been obtained for each channel. Thefiltering may shift each channel to baseband so that it can be processed as independent l/Qsignals. Alternatively, thefiltering may be performed at passband without shifting each channel to baseband.

Thiscomplex digital passband signal isin this particular embodiment then transformed into a real -valued digital passband signal by shifting the signal in frequency to be centered at half the sample rate Fs in step S102d' and then discarding, in step S102e, the imaginary component of thesignal illustrated in Figure 5c as previously has been discussed, which resultsin the real -valued digital passband signal illustrated with reference to Figure 5d to be provided for further transport in the DAS 100 as illustrated with step S103.

Again, as compared with a digital baseband representation, theresult isa signal with the same average data rate - i.e. half theamount of data, twice as often - but with a scheduling granularity that advantageously is half that of the baseband representation, si nee each of thedatasamplesisreal-valued and can be scheduled and routed independently.

Hence, as compared to thedigital baseband representation, where one real and one imaginary component must be processed in sequence for each data samplewith the corresponding delay, the processing delay for each data sampleof the real-valued digital passband signal when scheduling and serializing the data samples is reduced by 50%.

I n yet another embodiment, described with reference to theflowchart of Figure 8, the input to the source units 107, 108 of the DAS 100 is an RF signal having been mixed down to I F - but not to zero but to an appropriate passband frequency - with a non-quadrature mixer/demodulator . Again, the mixingand non-quadrature demodulation may alternatively be performed by a non-quadrature mixer/ demodulator 141, 142 located outside the source units 107, 108.

TheADCs 112, 113 will thus sample the analogue I F signal as a real -valued passband signal.

Again, oncethe I F signal has been digitized, the processing consists of separating thedifferent signalsthat makeup the composite input signal by frequency-selective filtering applied by the RX filters 114-117 as shown in step S102b' of Figure3b. Hence, each channel has a frequency-selective digital filter 114-117 whose bandwidth isadapted to the characteristics of the respective signal source, i.e. each filter istuned to the operati ng frequency of the corresponding RBS.

Thereafter, thesignalsof thefiltered channels are re- sampled or decimated with a lower sample rate Fsto adapt to the bandwidth of thesignalsof the filtered channels, as has been discussed with refer en ceto step S102b" of Figure3b.

Now, after thisfiltering, sincethe I F signal issampled asa real passband signal, a real-valued digital passband signal - as illustrated in Figure5d - has been obtained for each channel.

I n an alternative embodiment, described with referenceto theflowchart of Figure 9, again with an RF signal being received in step S101 and mixed down to I F and non-quadrature demodulated in step S102a'; oncethe I F signal has been digitized in step S102b, theprocessing consists of separating the different signals that makeup the composite input signal.

The digital real-valued passband signal attained after the ADC 112, 113 is processed for each channel with a digital quadrature demodulator in step S102b1 (implemented as a part of the RX filters 114-117) which shiftsthe real- valued passband signal for each channel to zero I F, by multiplying the received samples with a complex phasor with a frequency equal to the channel center frequency and wherethereal and imaginary components are 90 degrees out of phase, in a manner identical to theprocessing in an analogue quadrature demodulator. The frequency-selective RX filters 114-117 can then filter the real and imaginary components with low-passfilters adapted to the bandwidth of the wanted channel, as has been described in detail with referenceto step S102b' of Figure3b, si nee the wanted channel is centered at OHz. Thereafter, the data of thefiltered channel is re- sampled in step S102b" at a lower sample rate F _s adapted to the bandwidth of the data of thefiltered channel, the reduce the amount of data being passed on to the scheduler and serializer. The resulting baseband signal for each channel is then converted into apassband signal in step S103a, using the method described previously for digital baseband signals, prior to being passed to the scheduler and serializer for transmission over theoptical link.

In still another embodiment, theinput to the source units 107, 108 of the DAS 100 is an RF signal which is not mixed down but only passed through a non-quadrature demodulator (not shown). This demodulator may be located within or outside the source units 107, 108.

TheADCs 112, 113 will thus sample the analogue RF signal as a real -valued passband signal.

Again, oncethe I F signal has been digitized, the processing consists of separating the different signalsthat makeup the composite input signal by frequency-selective filtering. Hence, each channel has a frequency-selective digital filter 114-117 whose bandwidth is adapted to the characteristics of the respective signal source, i.e. each filter istuned to the operati ng frequency of the corresponding RBS.

Now, after thisfiltering, sincethe RF signal issampled asa real passband signal, a real-valued digital passband signal - as illustrated in Figure 6d - has been obtained for each channel.

In still afurther embodiment, theinput to the source units 107, 108 of the DAS 100 is an already digitized signal, i.e. a digital baseband signal.

In such scenario, thereisno need to demodulate or digitizetheinput signal, even though channel filtering still is necessary. The3-step approach already described with reference to Figures 6a-d hereinabove will thus be undertaken for the received digital baseband signal.. Further, as has been discussed, this may include a resampling of theinput signal to reduce the sample rate of each base station signal for efficient utilization of the digital link. As previously has been discussed, the DAS 100 illustrated in Figure 1 shows a downlink path, i.e. transmit path, from the base stations to wireless communication devices (not shown) such as smart phones, tablets, smart watches, gaming consoles, etc. I n an uplink path, i.e. receive path, the functionality of the remote units and source units would be reversed.

Even though the DAS 100 of Figure 1 isdivided into different functional entities such astheADCs 112, 113, the RX filters 114-117, the schedulers 118, 119, theserializers 120, 121, etc., a modern DASit typically implemented by means of a system of processing units executing computer programs to achieve the functionalities described herein.

With reference to Figure 10, the steps of the method performed by the DAS 100, and in particular the source units 107, 108, arein practice performed by a processing unit 30 (or system of processing units) embodied in theform of oneor more microprocessors arranged to execute a computer program 31 downloaded to a suitable storage medium 32 associated with the

microprocessor, such as a Random Access Memory (RAM), a Flash memory or a hard disk drive. The processing unit 30 is arranged to cause the DAS 100 to carry out the method according to embodiments when the appropriate computer program 31 comprising computer-executable instructions is downloaded to the storage medium 32 and executed by theprocessing unit 30. The storage medium 32 may also be a computer program product comprising the computer program 31. Alternatively, thecomputer program 31 may betransf erred to the storage medium 32 by means of a suitable computer program product, such asa Digital Versatile Disc (DVD) or a memory stick. Asafurther alternative, thecomputer program 31 may be downloaded to the storage medium 32 over a network. Theprocessing unit 30 may alternatively be embodied in theform of a digital signal processor (DSP), an application specific integrated circuit (ASIC), afield-programmable gate array (FPGA), a complex programmable logic device (CPLD), etc. Theinvention has mainly been described above with reference to a few embodiments. However, as is readily appreciated by a person skilled in the art, other embodiments than the ones disclosed above are equally possible within the scope of theinvention, as defined by the appended patent claims.