HK1091625A - Bit swapping for different interleaving depths - Google Patents

Description

Bit swapping for different interleaving depths

Technical Field

The invention relates to a method for bit swapping, wherein periodically I consecutive bits of a data packet comprising K bits are mapped onto interleaved bit positions of I different bursts, respectively, according to a predefined interleaving scheme and a selected interleaving depth I, comprising swapping the value of at least one bit associated with a respective first bit position m in the data packet with the value of a bit associated with a respective second bit position n in the data packet, wherein the respective second bit position n is selected such that n > m is maintained and the difference n-m is divisible by I.

Background

In most prior art communication systems, information originating from an information source is converted into bits, then source and channel encoded, interleaved, and modulated for transmission over a transmission medium, which may be the space between transmit and receive antennas or a wired connection such as a cable or fiber. Among the different modulation techniques, phase modulation has proven itself as a robust and efficient way of mapping information onto a carrier. In phase modulation, the phase of the carrier contains all the information of the transmitted bit.

EDGE (enhanced data GSM Environment), Global System for Mobile communicationsFaster versions of (GSM) wireless services are designed to transmit data at rates up to 384kbps and are capable of delivering multimedia and other broadband applications to mobile phone and computer users. The EDGE standard is built on the existing GSM standard. However, instead of using the Gaussian Minimum Shift Keying (GMSK) modulation technique originally standardized for GSM in EDGE, Phase Shift Keying (PSK) with eight possible symbol constellations (8-PSK) is used. Fig. 1 shows three consecutive bits (d) of an encoded, interleaved bit stream to be modulated_3i，d_3i+1，d_3i+2) Mapped onto one of eight possible 8-PSK symbols 1-1..1-8 on I/Q plane 2, where I represents the sequence number of the 8-PSK symbol, and where the I-axis and Q-axis represent the in-phase and quadrature components, respectively, of the modulated signal. All 8-PSK symbols 1-1..1-8 lie on a circle having the same radius, which differs only in its phase, which increases counterclockwise from the I axis.

When attempting to demodulate a received noisy 8-PSK signal in a symbol-wise manner to determine the associated three transmitted bits (d) for each 8-PSK symbol_3i，d_3i+1，d_3i+2) The phase of the received signal is determined by resolving the received signal into its in-phase and quadrature components, resulting in an estimated position (not shown) of the 8-PSK symbol on the I/Q plane 2. The estimated position is compared to the eight possible positions 1-1..1-8 shown in fig. 1 to determine which 8-PSK symbol was originally transmitted. However, if the symbol estimates are substantially different from the possible 8-PSK symbol positions, e.g., if the estimated symbols have a 22.5 ° phase and the transmitted 8-PSK symbols 1-1 have a 0 ° phase (corresponding to three bits (1, 1, 1)), it is difficult to decide whether the symbol 1-1 having a 0 ° phase or the symbol 1-2 having a 45 ° phase was originally transmitted. If it is erroneously decided that 8-PSK symbols 1-2 having a phase of 45 ° are transmitted, the demodulation results in bits (0, 1, 1) instead of the originally transmitted bits (1, 1, 1). Thus, one bit error occurs. It can be noted from fig. 1 that adjacent 8-PSK symbols always differ by only one bit position, so that the number of error bits is as low as possible when erroneously deciding the originally transmitted 8-PSK symbol as an adjacent 8-PSK symbol.However, even for errors originating from the detection of erroneous neighboring 8-PSK symbols instead of the correct 8-PSK symbols, the bit-triplet (d) is_3i，d_3i+1，d_3i+2) Also the error probability of (c) is different. Detection of adjacent 8-PSK symbols instead of the correct 8-PSK symbol may only result in a triplet (d) for 4 8-PSK symbols (1-1, 1-2, 1-5 and 1-6 at 0 °, 45 °, 180 ° and 225 °, respectively)_3i，d_3i+1，d_3i+2) The first position of the bit error, the triplet (d) is generated only for 4 8-PSK symbols (1-3, 1-4, 1-7 and 1-8 at 90 °, 135 °, 270 ° and 315 °, respectively)_3i，d_3i+1，d_3i+2) Bit error in the second position, and possibly a triplet (d) for all 8-PSK symbols 1-1..1-8_3i，d_3i+1，d_3i+2) Bit error in the third position. Thus, the triplet (d)_3i，d_3i+1，d_3i+2) Is more error prone than the first and second bit positions and is therefore referred to as the "weak bit" in the triplet. EDGE system (reference European Telecommunications Standardization Institute (ETSI) technical document 3gpp ts 45.003 v5.6.0(2000-06) allows multiplexing of multiple mobile stations onto a single uplink Packet Data Transport Channel (PDTCH) in order to control access of different mobile stations to the PDTCH, an Uplink State Flag (USF) is used which indicates whether the uplink channel is idle and, if not idle, to which mobile station it currently belongs, the USF has three bits, where "1" stands for "idle", the remaining 7 states can be used to identify which ms that is currently using the PDTCH-the USF flag is crucial for the EDGE system to function correctly and is therefore channel coded using a block code with code rate 1/12. One group of every 9 bits.

As shown in fig. 2, which represents the first block 3 of four blocks of a GSM/EDGE burst, each block comprises a total of 348 bits, with coded USF bits located at bit positions 168 to 173 and 176 to 178, respectively. The remaining bit positions in each block are filled with header and data bits that have been interleaved, encoded, and rate matched. The four blocks then form a burst of length 1392 bits.

For PDTCH, EDGE includes thirteen different Modulation and Coding Schemes (MCS). It is proposed to avoid in 8-PSK triplets (d) in MCS-5 and MCS-7 (in uplink and downlink)_3i，d_3i+1，d_3i+2) Transmits the coded USF bits at the third bit position to reduce the bit error rate of the USF. This principle is known as bit swapping. Bit swapping means that it will correspond to the 8-PSK triplet (d) in the burst to be treated as an 8-PSK triplet_3i，d_3i+1，d_3i+2) And the coded USF bits of the bit positions of the third bit transmission of (a) are not treated as 8-PSK triplets (d) corresponding to the data bits that have been interleaved coded and rate matched_3i，d_3i+1，d_3i+2) The bit positions of the third bit transmission are exchanged. Thus, the USF bit is only 8-PSK (d)_3i，d_3i+1，d_3i+2) This helps to reduce the bit error rate of the USF. As shown in fig. 2, the USF bits at positions 170, 173, and 176 (in 8-PSK symbols 56, 57, and 58, respectively, and shaded in gray in fig. 2) are swapped with the interleaved, coded, and rate-matched data bits at positions 150, 151, and 195 (not shown). It is readily seen that USF bit positions 168, 169, 171, 172, 177, 178 (unchanged) and 150, 151 and 195 (swapped) correspond to 8-PSK triplets (d) when the first bit of a burst has position 0_3i，d_3i+1，d_3i+2) The first and second error-prone bit positions because only bit position 3k-1, where k is 1..464, maps to the third error-prone bit position in the triplet.

The switching takes place in the transmitter. In the receiver, the demodulated bits from the received 8-PSK symbols are reverse swapped (de-swapped) based on knowledge of the swapping algorithm used at the transmitter. After the de-switching, the TFCI bit groups and the channel coded and rate matched data bit groups are channel decoded and then further processed.

Of course, when bit swapping is applied, the bit error rate of the data bits increases accordingly, since the error-prone third bit position in the 8-PSK triplet is now allocated to the data bits more frequently. However, when it is traded for the much more important USF reduced bit error rate, the bit error rate performance degradation of the data bits is more acceptable.

A similar situation occurs in the context of the flexible First Layer (FLO), which is a new type of physical layer, which is proposed for GSM/EDGE radio access networks (GERAN, refer to technical document 3GPP TR 45.902 v6.0 (2003-04) from ETSI-the main advantage of FLO is that the configuration of the physical layer (e.g. channel coding and interleaving) is specified at call setup-the GERAN physical layer provides one or more transport channels to the Medium Access Control (MAC) sublayer, through FLO, multiple transport channels can be multiplexed and sent on the same basic physical channel, coded composite transport channel (CCTrCH) at the same time, the configuration of transport channels, i.e. the number of input bits, channel coding, interleaving, etc. is called Transport Format (TF) -the transport format configuration is completely controlled by the Radio Access Network (RAN), and signaled MS. in the mobile station and the base transceiver station at call setup, the transport format is used to configure the encoder unit and the decoder unit. Only a limited number of TF combinations of different traffic channels (TrCH) are allowed. The effective combination is called a Transport Format Combination (TFC). In order to decode the received sequence, the receiver needs to know the active TFC for the radio packet. This information is conveyed in the Transport Format Combination Indicator (TFCI) field. This field is the header of the base layer. From the decoded TFCI value, the transport formats for the different transport channels are known and the real decoding can be started.

The size of the TFCI is limited to a maximum of 5 bits and a maximum of 32 different TFCs are allowed on the same basic physical subchannel. In other words, there are at most 32 different channel coding and/or multiplexing possibilities for a single connection at a time.

The TFSI is block coded and inserted at the beginning of a non-interleaved radio packet further comprising a multiplexed transport channel (CCTrCH). Each transport block of bits to be transmitted on the TrCH is appended with a Cyclic Redundancy Check (CRC) appended segment, channel coded, rate matched, and then multiplexed with other coded blocks to obtain a coded composite transport channel (CCTrCH). In a full-rate 8-PSK channel, the total length of the non-interleaved radio packet including TFCI and CCTrCH bits is 1392 bits. Before 8-PSK modulation, bits of a non-interleaved radio packet are subjected to block diagonal interleaving or block rectangular interleaving and interleaved onto I bursts, wherein I represents an interleaving depth. In the case of block rectangular interleaving, I bursts represent one radio packet. For example, on a full-rate 8-PSK channel, K1392 bits of a non-interleaved radio packet are interleaved into four bursts of J348 bits in size, which make up the 8-PSK modulated radio packet.

In the case of block angle interleaving, the bits of a non-interleaved radio packet comprising M-K/J non-interleaved bursts are interleaved onto I-2-M bursts of size J bits. However, the first I/2 bursts contain only bits at even bit positions, while the last I/2 bursts contain only bits at odd bit positions. Thus, the bits of this I burst must be combined with the bits of the other I bursts that result from interleaving the next non-interleaved radio packet into the I burst, resulting in two full radio packets from the two non-interleaved radio packets.

Due to the importance of TFCI for decoding received radio packets, it is desirable to improve the bit error rate of the TFCI. This can be achieved by bit swapping. However, in burst setup in the context of MSC-5 and 7 of EDGE, interleaving takes place before the USF bits, data and header bits are arranged in bursts and modulated, but in contrast, for FLO, the bits of TFCI and CCTrCH are jointly interleaved.

Thus, in the context of MSC-5 and 7, bit swapping can be performed directly after the burst is formed, since it is apparent which bit in the burst will be transmitted as a "weak bit" for 8-PSK modulation. In contrast, for FLO, joint interleaving of TFCI and CCTrCH produces I bursts, where it is apparent which bits will be transmitted as "weak bits". However, due to the joint interleaving of TFCI and CCTrCH, the position of the interleaved bits of TFCI in the radio packet depends on the interleaving scheme applied (block diagonal or block rectangular) and the different interleaving depths I (1, 2, 4, 8, 16) that are possible for full-rate, half-rate and possibly quarter-rate channels in the future, respectively. Thus, the bit swap must deal with different interleaving schemes and interleaving depths I.

Also, it is often desirable to only perform bit swapping between bits that are located in the same burst. This avoids affecting the temporal diversity, which is the main purpose of interleaving.

Disclosure of Invention

Thus, in view of the above, it is an object of the present invention to develop a simple method, system and computer program product for swapping higher and lower priority bits in a radio packet of a system that jointly interleaves the higher and lower priority bits and allows for different interleaving depths.

A method for bit swapping is proposed, wherein periodically I consecutive bits of a data packet comprising K bits are mapped onto interleaved bit positions of I different bursts, respectively, according to a predefined interleaving scheme and a selected interleaving depth I, comprising the steps of: exchanging at least one bit value associated with a respective first bit position m in the data packet with a bit value associated with a respective second bit position n in the data packet, wherein the respective second bit position n is selected such that n > m is maintained and the difference n-m is divisible by I.

For example, the bit at each bit position m may represent a high priority bit, while the bit at each bit position n represents a low priority bit. If the swapping is performed before the interleaving is performed, the swapping is done by swapping the values of the bits at two bit positions, i.e. for example, assigning the bit value at bit position n to the bit at bit position m and vice versa. The bit associated with each first bit position m in the data packet is then equal to the bit at bit position m and the bit associated with each second bit position n in the data packet is then equal to the bit at bit position n.

Interleaving is performed by periodically mapping I consecutive bits of the data packet onto interleaved bit positions of I different bursts, respectively. This may be done, for example, by periodically mapping the bits at bit positions K I + I (K0.., K/I-1, and I0.., I-1) in the data packet to different interleaved bit positions of burst I. The position of the interleaved bits within the burst is determined by the interleaving scheme, which may be, for example, block diagonal or block rectangular interleaving.

The exchange in each step is performed for bit positions m and N of the data packet, e.g. where N can be associated with m by a fixed mapping m-N + N, where N is a predefined natural number. To ensure that the swapped bits are in the same burst, the difference of the two bit positions, i.e., N-m-N, must be divisible by I because the period of interleaving is I.

Both switching and interleaving are performed at the transmitting site, which may be, for example, a mobile station or a base transceiver station of a mobile radio system, depending on the direction of transmission. At reception, a corresponding reverse swap (de-swap) and de-interleaving must be performed to rearrange the bits so that channel decoding can be performed. For the exchange, the value of at least one bit associated with a respective first bit position m in the data packet is de-exchanged with the value of a bit associated with a respective second bit position n in the data packet, wherein the respective second bit position n is selected such that n > m is maintained and the difference n-m is divisible by I. Thus, the switching step performed at the transmitter also describes the de-switching step performed at the receiver.

According to the invention, the swapping may be performed before, during or after interleaving at least I consecutive bits. Accordingly, at the receiver, the deinterleaving must be performed after, during or before the deinterleaving, respectively.

When the switching is performed before interleaving, the values of the bits at bit positions m and n in the data packet can be directly switched. The bit associated with each first bit position m in the data packet is the bit at bit position m in the data packet and the bit associated with each second bit position n is the bit at bit position n in the data packet. At the receiver, the de-swapping is then performed after de-interleaving and the bit values at the same positions m and n are de-swapped as during the bit swapping at the transmitter.

Conversely, when interleaving has been performed, the bit values that have now been mapped to interleaved bit positions in the I bursts must be swapped. These interleaved bit positions correspond one-to-one to the bit positions of the data packet by a predefined interleaving scheme and a selected interleaving depth. The bit associated with each first bit position m in the data packet is the bit at the interleaved bit position to which the bit at bit position m in the data packet is interleaved, and the bit associated with each second bit position n in the data packet is the bit at the interleaved bit position to which the bit at bit position n in the data packet is interleaved. It is thus possible to exchange the bit values after interleaving, i.e. by exchanging the bit values at interleaved bit positions in the burst, the bit values being selected, for example, according to their bit positions in the data packet. At the receiver, the deinterleaving must then be performed prior to deinterleaving based on the same exchanged interleaved bit positions as at the transmitter.

According to the invention, the selected interleaving depth I is preferably selected from a predefined set of interleaving depths I₁，..，I_RIs selected from where R has I for all R1_r＜＝I_maxAnd wherein preferably the respective second bit positions n are selected such that the difference n-m can be I_maxAnd (4) trimming.

If the interleaving depth I may have different values, it is preferable to require N not only to be the interleaving depth currently appliedDegree I is evenly divided, and the maximum interleaving depth I possible_maxAnd (4) trimming. Since the different possible interleaving depths are powers of 2, it is required that N must be able to be I_maxThe integer division ensures that for each possible I the exchanged bits are in the same burst, so that the temporal diversity is not affected.

According to the present invention, preferably at least one bit group is defined within the data packet, and wherein the step of swapping is performed only if the interleaved bit positions in the data packet to which the bits of the respective first bit positions m are mapped according to the predefined interleaving scheme and the selected interleaving depth I are characteristic interleaved bit positions, and if the bits at the respective first bit positions m in the data packet belong to the at least one bit group.

A group of bits in the data packet may represent high priority bits while the remaining bits in the data packet may represent low priority bits. Only the high priority bits that are mapped by interleaving onto interleaved bit positions that are considered characteristic interleaved bit positions in the burst are swapped.

According to the invention it is further preferred that the characteristic interleaved bit positions are characterized in that, depending on the modulation scheme, when bits are modulated, transmitted on a noisy channel and demodulated, the bits at these positions have a higher error probability than the bits at the remaining positions. Such "weak" bit positions occur, for example, on every third bit in 8-PSK modulation, and are also encountered in 16 Quadrature Amplitude Modulation (QAM) and 64 Quadrature Amplitude Modulation (QAM).

In accordance with the present invention, a typical interleaved bit position may be a position j within a burst that satisfies the condition that (j +1) is divisible by p, where p is a predetermined natural number greater than 0. For example, if the bits at the interleaved bit positions in a burst are to be 8-PSK modulated, every third bit will become a "weak bit", and for every position j in the burst, each "weak bit" is identified by the condition (j +1) mod p 0, where p 3, j ranges from 0 to the burst length minus 1.

According to the present invention, preferably, the bit group includes L first bits in the data packet, L being a predetermined number. The high priority bit may then represent a header added to the beginning of the data container.

According to the invention it is further preferred that the respective second bit position n is selected such that n-m ≧ L is retained. This condition ensures that the high priority bits in the coherent group of L bits at the beginning of the data packet are swapped with the low priority bits located in the rest of the data packet.

According to the present invention, preferably, the data packet comprises Transport Format Combination Identifier (TFCI) bits according to the GSM/EDGE radio access network (GERAN) Flexible Layer One (FLO) in L bit groups and coded combination transport channel (CCTrCH) bits according to the FLO of GERAN in the remaining K-L bits, wherein the K bits of the data packet are mapped onto interleaved bit positions in the burst according to one of the interleaving schemes standardized for the FLO of GERAN and one of the interleaving depths I, and wherein p-3 is maintained. The condition P-3 illustrates 8-PSK modulation in FLO of GERAN. In FLO in GERAN, different interleaving depths I, different full-rate and half-rate channels, and block diagonal and block rectangular interleaving schemes, selected from the set 4, 8, 16, are standardized.

According to the invention, it is preferred that at least two switching steps are carried out for the data packets, wherein the respective first bit position m in each step is different, wherein in at least one of the at least two steps the respective second bit position N is selected such that N ═ m + N is maintained, and wherein in at least one of the at least two steps the respective second bit position N is selected such that N ═ m + K-N is maintained, wherein N is a predetermined natural number. Alternate swaps may be implemented by a counter variable cpt, which is initialized to zero before the first swap and incremented by 1 after each swap. If (cpt mod 2 ═ 0), then m is exchanged for N ═ m + N, otherwise, m is exchanged for N ═ m + K-N. The basic principle of this alternating exchange with the CCTrCH starting bits and the CCTrCH ending bits is that the error protection of the CCTrCH is strongest at the beginning and end of the CCTrCH, respectively, so that the CCTrCH bit error rate is not excessively degraded by the CCTrCH bits transmitted in "weak" bit positions in order to let the TFCI bits be transmitted in less error prone bit positions.

Furthermore, a system for bit swapping is proposed, wherein I consecutive bits of a data packet comprising K bits are periodically mapped onto interleaved bit positions of I different bursts, respectively, according to a predefined interleaving scheme and a selected interleaving depth I, the system comprising processing means for swapping the value of at least one bit associated with a respective first bit position m in the data packet with the value of a bit associated with a respective second bit position n in the data packet, wherein the respective second bit position n is selected such that n > m is maintained and the difference n-m is divisible by I. The system may be embodied in a transmitter that performs the switching, such as a mobile station or base transceiver station in a mobile radio system, and/or in a receiver that performs the switching. The means for switching in the transmitter and for de-switching in the receiver are the same.

Furthermore, a computer program product directly loadable into the internal memory of a digital computer, comprising software code portions for performing the steps of the method as described above, when the product is run on a computer is proposed. A computer may already be present in the mobile station or base transceiver station for performing burst formation and interleaving.

Drawings

These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter. In the drawings, there is shown:

FIG. 1: symbol mapping of bit triplets to 8-PSK symbols,

FIG. 2: an example of the prior art of bit swapping in blocks of GSM/EDGE bursts,

FIG. 3: structure of radio packets in the GSM/EDGE Flexible Layer One (FLO),

FIG. 4: an interleaving table of block diagonal interleaving and interleaving depth I-8 is performed on the first 72 bits of a full-rate 8-PSK radio frame in FLO for GSM/EDGE,

FIG. 5: a block diagram of a system for bit swapping according to the present invention,

FIG. 6: a table showing the application of bit swapping according to the invention in full rate channels with interleaving depth I-8 in FLO for GSM/EDGE,

FIG. 7: a first comparison of the frame error rate obtained for FLO radio packets without bit swapping with the frame error rate obtained for FLO radio packets with bit swapping according to the present invention, and

FIG. 8: a second comparison of the frame error rate achieved by the FLO radio packet without bit swapping to the frame error rate achieved by the FLO radio packet with bit swapping according to the present invention.

Detailed Description

Fig. 3 shows the structure of a radio packet for a full rate 8-PSK channel in the GSM/EDGE Flexible Layer One (FLO). Each transport block, which contains the second layer binary data of the traffic channel (TrCH), is extended with a Cyclic Redundancy Check (CRC) appended segment to obtain a coded block, and then channel coding and rate matching are performed to obtain a radio frame. In fig. 3, only radio frames 4-0 and 4- (S-1) are shown corresponding to TrCH (0) and TrCH (S-1), respectively, where S represents the number of active trchs. For each radio packet to be transmitted, one radio frame 4-S, S-0, S-1 from each active TrCH is transmitted for TrCH multiplexing. These radio frames 4-s are serially multiplexed into a coded composite transport channel (CCTrCH). After adding TFCI bits 6 at the beginning of CCTrCH bits 7, a non-interleaved radio packet 8 results, in this example TFCI bits 6 are configured to 72 bits, and radio packet 8 comprises 1392 bits in total.

The TFCI 6 and CCTrCH bits 7 making up the non-interleaved radio packet 8 can now be block-rectangular interleaved to 4 bursts 9-0..9-3 or block-diagonal interleaved to 8 bursts 9-0..9-7, where in either case each burst size is J-348 bits. However, for block diagonal interleaving, I-8 bursts contain only bits at even (burst 9-0..9-3) or odd (burst 9-4..9-7) bit positions. In fig. 3, the case of block diagonal interleaving with I-8 is exemplarily shown. Along with the bits at the even and odd bit positions of the second non-interleaved burst, which are mapped onto the other 8 bursts, two full FLO radio packets 10 each 1392 bits in length, can be constructed. However, for simplicity, only one of the wireless packets 10 is shown in fig. 3. FLO radio packet 10 is then 8-PSK modulated.

In technical document 3GPP TR 45.902 v6.0.0, the interleaving that maps K bits of the non-interleaved radio packet 8 onto I bursts with J bits each can be defined by the following set of equations:

block diagonal interleaving with interleaving depth I:

D＝I

k＝0，1，2..K-1，

b＝k mod D。

rectangular interleaving of blocks with interleaving depth I:

D＝2·I

k＝0，1，2..K-1，

here, the function "int [ j ]" represents taking the smallest natural number smaller than j, while the modulo operator "imodj" returns the remainder of the i/j division.

These equations apply as follows: each bit at position K0.. K-1 is mapped onto an interleaved bit position J0.. J-1 within burst b 0.. I-1, K being the bit size of the non-interleaved radio packet/interleaved radio packet, including M non-interleaved bursts, and J being the bit size of the non-interleaved bursts and bursts.

Fig. 4 shows an interleaving table of the first 72 bits of a radio frame (TFCI bits) in FLO for GSM/EDGE, which obeys the above equation and sets parameter set K1392 and J348 (M4), i.e. a full-rate 8-PSK channel. Further, assume that block diagonal interleaving with an interleaving depth I of 8 is used.

The first column of fig. 4 represents the index k of the bits in the non-interleaved radio packet 8 before interleaving, the second column represents the assigned index j in the corresponding burst b after interleaving, and the third column represents the assigned burst index b after interleaving.

As can be seen from the first and third columns of fig. 4, the successive bits of the non-interleaved radio packet 8 are periodically mapped onto different bursts b, respectively. Furthermore, it can be seen that in the first four bursts b-0, 1, 2, 3, only even bit positions j are assigned values, while for the last four bursts b-4, 5, 6, 7, only odd bit positions are assigned values. Thus, 4 blocks of 348 bits per block are mapped from bits at the k-0.. 1392 bit positions onto 8 bursts of J-348 bit positions each, but each of these bursts comprises only bits at even or odd bit positions.

The wireless packet is then 8-PSK modulated, resulting in every third bit in the wireless packet being error prone than the other bits. Note that due to the fact that the bit size J of a burst is 348, which can be divided exactly by 3, the position of a bit in the burst is sufficient to decide whether this bit will be located in the third bit position of the corresponding 8-PSK symbol, i.e. the absolute bit position in the radio packet does not need to be known. The TFCI bit to be transmitted as the third bit can be identified by searching for an index j corresponding to the top k 0..71 bit position and satisfying the condition (j +1) mod 3 0. This is the case for bit positions k of 1, 5, 10, 14, 16, 19, 20, 23, 25, 29, 34, 38, 40, 43, 44, 47, 49, 53, 58, 62, 64, 67, 68 and 71 (see fig. 6). To improve the TFCI bit (or frame) error rate, according to the invention, the bits at these bit positions are now swapped with the bits from the CCTrCH, i.e. bit K is swapped with bit K + N when cpt is even and bit K is swapped with bit K + (K-N) when cpt is odd, where N is a predetermined natural number, cpt is a counter, which is initialized to zero when the swapping process starts and incremented by 1 after each swapping operation. Note that swapping bits K alternately with bits K + N and K + (K-N) only guarantees that TFCI bits are swapped with bits at the beginning and end of the CCTrCH, where error protection is strongest. Optionally, bits k are swapped with bits k + N regardless of the value of cpt, which simplifies the process, but may even further increase the bit error rate of the CCTrCH.

Fig. 5 shows a block diagram of a system for bit swapping according to the present invention. The system comprises a burst memory 11 with Interfaces (IF)12, 13 and 14, a processor 15 for controlling the interfaces 12, 13 and 14, and a look-up table (LUT) 16. The processor controls burst-wise storage of bursts of interleaved radio packets 10 from a previous interleaving stage via IF 12 into burst memory 11, which burst memory 11 may be implemented as RAM. When storing the burst in the RAM, the processor swaps the bits of the stored burst through the interface 13 according to the swapping information of the current interleaving scheme and interleaving depth I contained in the LUT 16. Finally, the processor triggers a read of the burst from the memory 11 via the interface 14. The switched bursts 17 are then forwarded to the modulator stage.

The above description assumes that the switching is performed at the transmitter. The same settings can be used for de-swapping at the receiver, where incoming bursts are stored in burst memory 11, de-swapped according to the same LUT 16 as in the swapping case, and then forwarded to the de-interleaving stage.

Fig. 6 shows a table indicating which bits of the TFCI are exchanged with bits from the CCTrCH according to the present invention. Such a table may be stored in the LUT 16 of the system for bit swapping as shown in fig. 5. For example, according to the invention, N-80 is selected, which may be denoted by I_max16 and is greater than or equal to the size of the TFCI (L72 bits). Note that N-80 is chosen so that the same bitswap process can be applied to all interleaving schemes and interleaving depths possible within FLO, i.e., taken from the set {4, 8, 16} and R-3. Otherwise, it is toIn the case of diagonal block interleaving of I-8, which currently aims at bit swapping at the beginning and end of the CCTrCH, it is sufficient to choose N-72, which is larger than or equal to the size L-72 of the TFCI and which can be divided exactly by I-8.

The first column represents the index k of the bit position in the non-interleaved radio packet 8. The second column represents the absolute position k' ═ b × J + J of the CCTrCH bits exchanged with the TFCI bits in the non-interleaved radio packet 8. The third column gives the burst index b in which the swap takes place. The last two columns contain bit position j (within burst b) to which the bit at position k is interleaved, and bit position j '(within burst b) to which the bit at position k' is interleaved. For example, the TFCI bit with position k equal to 25 is interleaved to bit position j equal to 14 of burst b equal to 1, which satisfies (14+1) mod 3 equal to 0, i.e., this TFCI bit will be transmitted at the third bit position of the 8-PSK symbol, and thus must be swapped. Then, the switching is performed between the bit at position k 25 and the bit at position k +80 105 (beginning of CCTrCH). The bit at position 105 is interleaved at j' 198 within burst b 1, which does not satisfy (198+1) mod 3 0. Note that the bits within the same burst b are always swapped to preserve temporal diversity.

Fig. 7 shows a first comparison of the frame error rate obtained for FLO radio packets without bit swapping and for FLO radio packets with bit swapping according to the present invention. FLO is configured to carry a 4.75kbps adaptive multi-rate coded (AMR) call over an 8-PSK channel with a 5-bit TFCI (72-bit code). The TU3iFH channel is used as a channel model. Fig. 7 shows the Frame Error Rate (FER) of radio packets comprising TFCI and CCTrCH and the FER of radio packets comprising only TFCI, both with (dashed) and without (solid) bit swapping according to the invention, where FER is a function of carrier-to-interference ratio (C/I) in dB. For the FER of TFCI, a gain of 1.3dB (FER 0.01) can be obtained by applying bit swapping according to the present invention. The combined FER of CCTrCH and TFCI shows a gain of 0.2 dB.

Fig. 8 shows a second comparison of the frame error rate obtained for FLO radio packets without bit swapping and for FLO radio packets with bit swapping according to the present invention. Since more weak bits are used for the CCTrCH as a result of the bit swap, this will cause some performance loss when many bits are swapped and when the coding rate of the CCTrCH is high (which corresponds to low error protection). To estimate the loss, FLO is configured to carry a 12.2kbps AMR call on the same channel as fig. 7. Fig. 8 shows the results of expressing the combined FER of TFCI and CCTrCH with (dashed line) and without (solid line) bit swapping as a function of C/I in dB. The performance degradation caused by bit swapping in this case is negligible.

The invention has been described above by means of preferred embodiments. It should be noted that it will be apparent to those skilled in the art that: alternative methods and variations exist and may be implemented without departing from the scope and spirit of the appended claims, e.g. bit swapping may be performed during the interleaving step and different interleaving schemes may be applied, particularly with respect to the arrangement of interleaved bits within a burst. The scope of the invention is in no way limited to 8-PSK modulation or to GSM/EDGE systems. It may also be applied in spread spectrum or Orthogonal Frequency Division Multiplexing (OFDM) systems, for example.

Claims (12)

1. A method for bit swapping, wherein periodically I consecutive bits of a data packet comprising K bits are mapped onto interleaved bit positions of I different bursts, respectively, according to a predefined interleaving scheme and a selected interleaving depth I, comprising the steps of:

exchanging at least one bit value associated with a respective first bit position m in said data packet with a bit value associated with a respective second bit position n in said data packet, wherein said respective second bit position n is selected such that n > m is maintained and the difference n-m is divisible by I.

2. The method of claim 1, wherein the swapping is performed before, during, or after the interleaving of the at least I consecutive bits.

3. The method of claim 1, wherein the selected interleaving depth I is from a predefined set of interleaving depths { I }₁，..，I_RIs selected from where R has I for all R1_r＜＝I_maxAnd wherein said respective second bit positions n are selected such that the difference n-m can be I_maxAnd (4) trimming.

4. The method of claim 1, wherein at least one group of bits is defined in the data packet, and wherein the exchanging step is performed only if:

if the bit at said respective first bit position m in said data packet is an interleaved bit position which is characteristic for the interleaved bit position to which said predefined interleaving scheme and said selected interleaving depth I are mapped, and

if said bit at said respective first bit position m in said data packet belongs to said at least one bit group.

5. The method of claim 4, wherein bits at the characteristic interleaved bit positions have a higher probability of error than bits at the remaining positions when the bits are modulated, transmitted over a noisy channel, and demodulated based on a modulation scheme.

6. The method of claim 4 wherein the characteristic interleaved bit position is a position j in the burst that satisfies the condition that (j +1) is divisible by p, where p is a predetermined natural number greater than 0.

7. The method of claim 4, wherein the group of bits includes L first bits in the data packet, where L is a predetermined number.

8. The method of claim 7, wherein the respective second bit positions n are selected such that n-m ≧ L is maintained.

9. The method of claim 8, wherein the data packet comprises Transport Format Combination Identifier (TFCI) bits in accordance with a Flexible Layer One (FLO) of a GSM/EDGE radio access network (GERAN) in the set of L bits and a code combined transport channel (CCTrCH) bit in accordance with the FLO of the GERAN in a remaining K-L bits, wherein K bits of the data packet are mapped onto the interleaved bit positions in the burst according to one of an interleaving scheme and one interleaving depth I standardized for the FLO of the GERAN, and wherein p is maintained at 3.

10. The method according to claim 9, wherein the exchanging step is performed at least twice for the data packet, wherein the respective first bit position m in each step is different, wherein in at least one of the at least two steps the respective second bit position N is selected such that N ═ m + N is maintained, and wherein in at least one of the at least two steps the respective second bit position N is selected such that N ═ m + K-N is maintained, wherein N is a predetermined number.

11. A system for bit swapping, wherein periodically I consecutive bits of a data packet comprising K bits are mapped onto interleaved bit positions of I different bursts, respectively, according to a predefined interleaving scheme and a selected interleaving depth I, comprising:

-processing means for exchanging at least one bit value associated with a respective first bit position m in said data packet with a bit value associated with a respective second bit position n in said data packet, wherein said respective second bit position n is selected such that n > m is maintained and the difference n-m is divisible by I.

12. A computer program product directly loadable into the internal memory of a digital computer, comprising software code portions for performing the steps of claim 1 when said product is run on a computer.