GB2361851A

GB2361851A - Blind data rate determination

Info

Publication number: GB2361851A
Application number: GB0010306A
Authority: GB
Inventors: Vagan Shakhgildian
Original assignee: Motorola Inc
Current assignee: Motorola Solutions Inc
Priority date: 2000-04-28
Filing date: 2000-04-28
Publication date: 2001-10-31
Also published as: GB0010306D0

Abstract

In a cellular communications system, a method of detecting the data bit rate in a received data frame comprises splitting the frame into two portions at the transmitter. The first portion is convolutionally encoded at a known maximum rate by repeating data bits within the first portion. The second portion is encoded at a lower optimum rate. The size of the first portion is the minimum necessary to reliably estimate the data rate. At the receiver the first portion is decoded by a Viterbi decoder using the trellis corresponding to the maximum coding rate. Due to bit repetition the set of possible states of the trellis will be a subset of the maximum rate trellis. This subset corresponds to a particular data rate and is analysed therefore to determine the data rate. The second portion of the data frame is then decoded at the lower optimum rate. Thus only a single decoder is required.

Description

2361851 BLIND DATA RATE DETERMINATION This invention relates to the

automatic detection of transmissionlencoding data rates, and more specifically to a method of automatically detecting the data rate of received data in a cellular communications system.

Multi-rate data transmission from mobile users is at the present time used within digital cellular communications systems, and there is a requirement in the future for such systems to be able to accommodate continuous data rates ranging from 8 to 2048 kbps.

For a receiver in a mobile handset or station (MS), or base transceiver station (BTS) to detect successfully an incoming transmission, the data rate of transmission (the bit rate) of the signal must be known. However, the data rate of transmission by a system user can change dynamically on a frame by frame basis.

Data rate information can be obtained through in-band or out-of-band signalling.

When in-band signalling is used, a number of control bits in each frame may be allocated to the bit rate information. As an example, for a bit rate range of 8 to 2048 kbps, 8 bits will be required to carry the bit rate information per frame. These 8 bits need to be protected by coding more powerful than normal traffic channel coding.

This results in a noticeable amount of system resources, such as bandwidth, being allocated for this purpose. Obviously, this is undesirable. The less resources tied up, the better the performance of the system will be.

An alternative to the above method of transmitting the data rate value within the signal is for the receiver to detect automatically the user bit rate from the incoming signal. This is commonly known as "blind rate determination" (BRD). BRID may be implemented in cellular communications systems at the expense of more complex receivers and a slight degradation in performance with regard to the above described method. This cost is due to the fact that BIRD makes no use of a priori information about the user bit rate.

2 One known method of performing BIRD consists of modifying the data coding rate so that the resulting coded data stays constant when the bit rate of the channel encoder input changes. The decoder in the receiver runs X concurrent decoding algorithms, where X represents the number of possible data rates and thus the number of 5 possible coding rates. All X decoded streams of data are tested, for example using a CRC (correlation) check, and the one with the best result is selected. Obviously the definition of the best result is dependent upon the testing criteria.

It thus becomes clear that in order to carry out the above technique, i.e. the decoding and testing of X streams of data, parallel processing is required. Such parallel processing requires an X times increase in the signal processing workload over a single process. This places a heavy burden on system resources and is a great drawback of this technique. These processing requirements are the main drawback to the use of BIRD in cellular communications systems. Thus there exists a requirement for significant processing reductions in the BIRD technique.

The present invention aims to address the above disadvantage.

The present invention provides, as claimed in the appendant claims, a method of, and apparatus for, automatically detecting the data bit rate used in a data frame received in a receiver of a cellular communications system, wherein signal repetition is used in a portion of the frame to define the data bit rate.

This invention addresses the problems set out above with regard to currently known methods of blind data rate determination, by providing a method based on variable convolutional encoding at the transmitter and a simple rate estimation procedure at the receiver. As such there is a requirement for only a single decoder, thus reducing the processing requirements of the receiver over the known method.

An exemplary embodiment of the invention will now be described with reference to the accompanying drawings, in which:

Figure 1 is a block diagram illustrating communication between a system transmitter and a system receiver of a cellular communications system in accordance with an embodiment of the invention; 3 Figure 2a is a flow diagram illustrating the operation of an embodiment of the method of the present invention; Figure 2b depicts a single data frame split in accordance with the present invention; Figure 3a is an example of a code trellis relating to the present invention; Figure 3b is an example of a reduced code trellis relating to the present invention; and Figure 4 is a flow diagram of the method of data rate determination according to an embodiment of the present invention.

The preferred embodiment of the present invention is now described with reference to the accompanying drawings as detailed above.

Referring to Figure 1 of the drawings, a cellular communications system transmitter 2 is provided with an encoder 4, a modulator 6 and transmission means 8. Similarly, a receiver 10, is provided with receiving means 12, a demodulator 14, data rate estimation means 16 and a decoder 18. Clearly, both the transmitter 2 and the receiver 10 may be in a base transceiver station or a mobile station or otherwise.

In operation, the data to be transmitted is encoded in the encoder 4, modulated in the modulator 6, and transmitted by the transmission means 8 to the receiving means 12 of receiver 10. The received data is demodulated by the demodulator 14 and an estimate is made of the bit rate by the data rate estimation means 16. Finally, the data is decoded at the correct data rate by the decoder 18.

With regard to Figure 2, the method of an embodiment of the present invention is now described. The method described relates to the case when the data transmission bit rate of a system user changes dynamically in powers of two.

The present invention generally follows the procedure commonly used within the art.

However, at the start of the process the data to be transmitted, represented by data frame 220 shown in figure 2b, is split in step 202, i.e. each data frame 220 is split into 4 two portions 221,222. The length of the first 221 of the two portions is denoted as L, as seen in Figure 2b. The length L is an indication of the number of bits within the first section 221 of the data frame 220. L depends upon the decoding delay used in step 209. The decoding delay is required to deliver a sufficiently low bit error rate (BER).

As mentioned above, the standard procedure of encoding data for transmission, modulating the data and transmitting the data is adhered to in the broad sense. However, once a data frame 220 has been split into two portions 221,222 in stage 202, the encoding stage 203 operates in a particular fashion. This will be best illustrated by example.

In order to encode data for transmission, for the purpose of this example, a forward error correction (FEC) convolutional coder is employed. In order to transmit the highest data rate, a rate of 1/k and a constraint length (i.e. the length of the shift register used to encode the data) of u are used. Three stages of a trellis diagram exemplary of such an encoder with a constraint length of 3 and k = 2 are depicted in Figure 3a.

A trellis diagram plots the state of the encoder along the vertical (y) axis against time on the horizontal (x) axis. The diagram indicates the transitions between the states occupied by the encoder at successive times. As is obvious therefrom, the encoder of this example is a three bit convolutional encoder.

When the transmission data rate drops below the highest rate, the drop in rate being signified by N, the encoder is still clocked at the same rate at which it was being clocked when the rate of data input was at its maximum. As such, when the transmission data rate is slower than the encoding unit, the last data symbol received in the encoder is repeated until a new symbol is received. Thus each symbol (i.e. a data symbol (i.e. a data symbol) as yet unencoded, is repeated N times at the 1/k rate encoder's input.

In view of the change in the data input rate, the encoder may now be seen as having an encoding rate of 1/(kN). Such an encoding rate will produce a trellis, shown in Figure 3b, which is a subset of that shown in Figure 3a. The number of used states in the code trellis of Figure 3b is 4. This means that the equivalent constraint length of the ISR code is 2, which is less than the constraint length of the original code which was 3.

The first portion 221 of the data frame 220 is encoded using the above ISR techniques. The second portion 222 of the data frame 220, from the end of the first portion 221 to the end of the second portion 222, is encoded with the optimum code.

The optimum code has a rate of (IlkN). This is one of the best known codes with this rate, such that it provides a good link performance. In the case where N = 1, it may be the same as the original code with a rate of (l /k).

At this point it becomes pertinent to detail why the data frame 220 is split into two sections and encoded using two distinct rates rather than simply encoding the entire data frame 220 using the method and rates detailed above.

The code produced by the encoder operating under the ISR scheme detailed above has a constraint length which is reduced with regard to that of the original convolutional code, i.e. that produced by the encoder operating at the maximum data rate. As such, the performance of the ISR scheme will be inferior to that of the scheme operating at the optimum encoding rate, because the encoding will be carried out using a smaller number of bits. In addition, the difference in performance increases as the input data rate reduces.

Finally, the splitting of data frames 220 reduces problems that may be experienced in decoding when using ISR decoding procedures. These procedures, although similar in some ways, have significant differences, such as path tracing for different encoding rates.

In view of the above detailed shortcomings, it is considered beneficial to split the data frames 220 for transmission into two portions 221,222. The first portion 221 is encoded according to the ISR scheme and the remainder is encoded at the optimum rate. The number of bits within the period L of the data frame is kept to the minimum required to estimate reliably the data rate in order to reduce as much as possible the degradation of the signal experienced relative to the potential performance for a given data rate.

6 Therefore, in step 203, the first portion 221 of the data frame 220 is encoded following the ISR scheme in order to produce a coded frame segment coded at a known rate i.e. the maximum rate (l/k). In step 204 the remainder of the data frame 220 is encoded at the optimum rate, i.e. generally the rate at which the input data is being 5 input, specifically the data rate of transmission of the system user.

Following step 204, the data to be transmitted is modulated and transmitted according to procedures usual in this field of technology. Similarly, the transmitted data is received and demodulated at steps 207 and 208 respectively.

The next step of the procedure is the estimation of the data rate (function box 209). During the first portion 221 of a data frame 220, the information bits which are contained within that section of the frame are decoded by a Viterbi decoder using the trellis corresponding to the maximum encoding rate (11Nk) where N = 1, as shown in Figures 3a and 3b. The delay in decoding received data is expressed as L, bits, i.e. the number of bits received that are awaiting decoding. As such, the decoded path, after receipt of L bits (the number of bits in the first section (221) of data frame (220) signified by the length L) is given by:

Lo = (L - L,) Where Lo represents the number of states in the decoded path. As stated previously, the value of L depends upon the decoding delay of the decoder employed. For a 1/k encoding rate, the decoding delay will typically be close to 5u, where x) is the constraint length of the convolutional encoder.

With regard to an encoding/decoding trellis, an example of which is seen in Figures 3a and 3b, there are:

LO x 2' states in LO stages of the trellis. This set of states can be partitioned into M enclosed subsets. The largest subset corresponds to the case where N = 1 and is identical to the whole set. As such, M indicates the number of possible data rates and each of the subsets corresponds to a particular data rate, i.e. a specific value of 1/(kN).

7 Using again the example above, the trellis of Figures 3a and 3b indicates the scenario where M = 2, i.e. the scenario where there are two subsets of states. The first subset is indicated in the right-most column of Figure 3a and includes all eight states. This corresponds to N = 1. The second subset is indicated in the right-most column of Figure 3b. This subset consists of the four shaded states and corresponds to N = 2, i.e. a code rate of 114.

After the bits within the first portion 221 of the data frame 220 have been decoded using the maximum decoding rate (l/k), the decoder analyses the LO states of the decoded path with respect to the appropriate trellis in an attempt to ascertain the subset to which the path belongs, thereby determining the data rate. Once the data rate has been determined, the remainder of the data frame is decoded at the optimum rate as indicated by function box 210.

The method of determining the data rate indicated at function box 209 is now described with reference to Figure 4. The method, as stated above, attempts to isolate the subset of states to which the states present in the decoded path belong, i.e. which progression of states in the trellis diagram most closely resembles the states present in the decoded path. The method starts by assessing the subset with the maximum value of N, i.e. the maximum drop in data rate. All the subsets of states are preferably recorded in order of decreasing value of N in the decoder memory.

Function box 401 presents the first step taken in determining to which subset of states the states present in the decoded path belong. An assessment of the number of states from the decoded path which belong to the first subset (where N is a maximum) is made. If that number is greater than or equal to the threshold T, set to a suitable level for the purposes of this method, then the rate of encoding is equal to Ri = R.,,x Nmax where Rmax is the maximum data rate and Nmax is the maximum drop in data rate.

T is an indicator assigned as an arbitrary adjustable threshold which may differ for the different stages of the method detailed below.

8 However, should the threshold T not be equalled nor exceeded, the assessment will be repeated in step 402 for the second subset. If the threshold T which may or may not remain the same for the second subset, is equal or exceeded, then the data rate 5 corresponds to that represented by the second subset, i.e.

Ri = Rmax Nniax - 1 Again, should the threshold T not be equalled nor exceeded, the next subset will be assessed. This continues until the subset corresponding to the decoded path is isolated, up until the Mth subset. Should analysis with respect to the M th subset be carried out, then the data rate will be represented by:

Ri = Rmax =R (N,,x - M) In order to remove dependance of the performance of the above method on the data being transmitted, the data are scrambled prior to being encoded in the transmitter.

Such a procedure ensures that the probability of data symbols being repeated at the encoder input, because of a particular data pattern rather than a reduction in the data rate, is extremely low. Additionally, the data are interleaved prior to being scrambled, in order to counteract degradation in the performance of the encoding at the beginning of a frame. Any degradation across the first portion (221) of the data frame (220) will be averaged across the whole data frame (220) when it is de-interleaved.

For the implementation of the above method, a small increase in complexity of the receiver is required. More specifically, an element of memory and increased processing facilities will be required over an apparatus used in non-BRID systems.

In addition to the foregoing, the present invention provides an apparatus comprising the means necessary to carry out the stages of the method described above.

As will be appreciated, the proposed method of "blind" data rate determination is based on variable convolutional encoding at the transmission end of the link and simple rate estimation procedure at the receiving end. The method avoids having parallel Viterbi decoding of several codes, which is characteristic of known techniques, and therefore has significant advantages in receiver complexity over existing methods.

1 . i 9 It will of course be understood that the present invention has been described by way of example, and that modifications of detail can be made within the scope of the invention.

Claims

1 ' A method of automatically detecting the data bit rate used in a data frame received in a receiver of a cellular communications system, wherein signal repetition is used in a portion of the frame to define the data bit rate.

2. A method according to claim 1, comprising the steps of:

splitting each data frame; encoding each data frame; transmitting and receiving each data frame; estimating the data rate of each data frame; and decoding the data.

3. A method according to claim 2, further comprising the steps of: 15 modulating the data frame; and demodulating the data frame.

4. A method according to any preceding claim, wherein each data frame is split into two sections.

5. A method according to claim 4, wherein the first section comprises the number of data bits necessary to estimate the data rate of that data frame.

6. A method according to claim 5, wherein the number of data bits are dependent upon the decoding delay of a decoder utilised in decoding the data frame.

7. A method according to any of claims 4 to 6, wherein the step of encoding each data frame comprises: encoding the first section of a data frame at a first code rate; and 30 encoding the second section of a data frame at a second code rate.

8. A method according to claim 7, wherein the first code rate is the maximum code rate.

9. A method according to claim 7 or claim 8, wherein the second code rate is the optimum code rate.

10. A method according to any of claims 4 to 9, wherein the step of estimating the data rate comprises: decoding the first section of each data frame at the maximum code rate; and analysing the decoded section.

11. A method according to claim 10, wherein the step of analysing the decoded section comprises: comparing the states within the decoded section with each subset of states generated by the encoder; and establishing whether the comparison yields a match to a preset threshold.

12. A method according to claim 11, wherein if the threshold is equalled or exceeded for a particular subset, the data rate corresponding to that subset is the rate sought.

13. A method according to claim 11 or claim 12, wherein the subsets are analysed in order of increasing code rates.

14. A method according to any of claims 4 to 13, wherein the second section of each data frame is decoded at the estimated code rate.

15. An apparatus for automatically detecting the data bit rate used in a data frame received in a receiver of a cellular communications system, wherein the receiver utilises only a single decoder.

16. An apparatus according to claim 15, comprising:

means for splitting each data frame; means for encoding each data frame; means for transmitting and for receiving encoded split data frames; means for estimating the data rate of each received data frame; and means for decoding each data frame.

--- 12

17. A method of automatically detecting the data bit rate used in a data frame received in a receiver of a cellular communications system substantially as hereinbefore, described with reference to the attached figures.

18. An apparatus for automatically detecting the data rate used in a data frame received in a receiver of a cellular communications system substantially as hereinbefore described with reference to the attached figures.