Classifications

• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L5/00—Arrangements affording multiple use of the transmission path
  • H04L5/003—Arrangements for allocating sub-channels of the transmission path
  • H04L5/0044—Arrangements for allocating sub-channels of the transmission path allocation of payload
  • H04L5/0046—Determination of how many bits are transmitted on different sub-channels
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L5/00—Arrangements affording multiple use of the transmission path
  • H04L5/003—Arrangements for allocating sub-channels of the transmission path
  • H04L5/0058—Allocation criteria
  • H04L5/006—Quality of the received signal, e.g. BER, SNR, water filling
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L1/00—Arrangements for detecting or preventing errors in the information received
  • H04L2001/0098—Unequal error protection
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L27/00—Modulated-carrier systems
  • H04L27/26—Systems using multi-frequency codes
  • H04L27/2601—Multicarrier modulation systems
  • H04L27/2602—Signal structure
  • H04L27/2604—Multiresolution systems
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L5/00—Arrangements affording multiple use of the transmission path
  • H04L5/0001—Arrangements for dividing the transmission path
  • H04L5/0003—Two-dimensional division
  • H04L5/0005—Time-frequency
  • H04L5/0007—Time-frequency the frequencies being orthogonal, e.g. OFDM(A), DMT

Definitions

• UEP codes are a well-known tool to protect data according to quality requirements or importance levels.
• RCPC rate- compatible punctured convolutional codes
• DVB-t denoted as “hierarchical modulation 1 , where the constellation points are not equally distributed (Fig. 3) .
• DVB-t are based on multicarrier modulation, denoted by OFDM (Orthogonal Frequency Division Multiplex) in wireless and DMT (Discrete MultiTone) in wireline transmission. Knowing the channel characteristics at the transmitter, multicarrier transmission offers the flexibility to adapt the modulation format on every carrier according to the actual signal-to-noise ratio. This adaptation is called v bit-loading ' in wireline and ' adaptive modulation' in wireless communications. Bit-loading is implemented in current ADSL and VDSL transmission systems.
• the bit-loading (adaptive modulation) is modified such that the number of bits (constellation size) is reduced for higher protection and increased for lower protection relative to the equal-protection case.
• margins which define the protection levels in terms of the signal-to-noise ratio.
• This is a standard formlation of additional protection, also in the equal-protection case where an additional protection against unpredictable additional disturbances is required.
• xDSL standards use a constant margin of 6 dB.
• the said invention introduces a non- mtuitive, but robust scheme, where the sensitive data is placed on carriers with a low signal-to-noise ratio.
• This offers robustness again non-stationary noise events that were not considered during the bit-allocation process.
• This makes use of the fact that the chance to deteriorate high signal-to-noise ratios is very high, whereas already low signal-to-noise ratios will not change much when unexpected additional noise courses affect transmission.
• Such typical noise sources are seen in impulse noise (wireline) , but also in sudded distortions in wireless transmission.
• the objective of the invention is realized by a method for unequal- error protection bit loading for multicarrier transmission to adapt the error probability and the signal-to-noise ratio margin to requirements and the sensitivity of the data, especially source- encoded data, e.g., coming from a video or audio encoder, using a bitallocation algorithm, where different margins and margin separations corresponding to different bit-error ratios modify the number of bits and selecting the modulation alphabet (signal constellation) such that the signal-to-noise ratios will show these chosen margin separations for the bits of different sensitivities and characterized in that the channel capacity or approximations thereof are used for determining the bit load modifying the signal-to-noise ratio by the said different margins with a predefined margin separation such that the corresponding bit-error ratios relative to the signal-to-noise ratio will show these predefined margin separations.
• the meaning of "margins” is the meaning according the ADSL standard G.992.1.
• Margin separation is the distance of different margins of the classes
• bit number results from quantizing (e.g. rounding) the real bit load of the k th carrier in the j th
• a method for unequal error protection bit loading characterized in that the data is subdivided into sensitivity classes Mj and the carriers are sorted according to the signal-to-noise ratios at these carrier frequencies and the carrier groups belonging to a certain protection class are specified as consecutive entries in this sorted carrier list, where this ordering and grouping is equivalent to introducing thresholds in the signal-to-noise ratio where the number of thresholds is one less than the number of different sensitivity classes.
• SNR margins are allocated to the carrier groups such that they decrease with the signal-to-noise ratio, i.e., the higher the signal-to-noise ratio, the lower the chosen margin moving the most sensitive data to the carriers with the worst signal-to-noise ratio.
• bit-loading is not performed over channels represented by the different carriers of a multicarrier system, but alternatively or additionally over spacial or temporal channels, including space-time, space-frequency, space- time-frequency, and MIMO (multiple-input multiple-output) systems.
• the UEP bit loading is performed at the transmitter on the basis of signal-to-noise ratios measured at the receiver requiring a duplex transmission channel and a demapping or decoding at the receiver according to the signal alphabet resulting from the UEP bit loading process.
• the apparatus for unequal error protection bit loading is adapted for carrying out a method according to any one of claims 1 to 14.
• a computer program comprising program code means for causing a computer to perform the steps of one of the methods as claimed in any of the claims 1 to Claim 15 when said computer program is carried out on a computer.
• the invention relates to a method for the transmission of data of different sensitivity in accordance with Claim 1
• UDP unequal error protection
• e physical layer
• the unequal-protection bit loading can also be applied to other channel scenarios, e g , in the form of space-time, space-frequency, space-time-frequency, and MIMO (multiple-input multiple-output) constellations
• Campello alg This is a lower complex alternative to Hughes-Hartogs applying some carrier grouping (Campello, J , Practical Bit Loading for DMT, proc ICC, Vancouver, 1999, pp 801-805)
• George- Amrani alg. This is a greedy algorithm placing bits where it increases the overall bit- or symbol-error rate least. (George, Y., Amrani, O., Bit loading algorithms for OFDM, proc. ISIT 2004, June 27, - July 2, 2004, p. 391)
• BRMP bit- rate maximization problem
• the BRMP may be converted other problems as follows: the designed symbol-error rate is exceeded: the algorithm may reduce the number of bits to achieve a lower SER. This changes the algorithm to be a probability of error minimization problem (PEMP) . the achieved probability of error is less than the designed value: the algorithm is re-iterated to reduce the power until the SER is achieved, keeping the number of bits constant. This is equivalent to a power minimization problem (PMP).
• PEMP probability of error minimization problem
• the invention according to claims 1 and 2 can be regarded as a generalization of the algorithm by Chow et al.. It computes the bit- load based on Shannon's capacity formula by introducing different noise margins ⁇ 3 for the N 9 groups. ⁇ j is computed iteratively to fulfill the required number of bits T 3 for the set of subcarriers M j .
• the power P k is allocated based on the SER of class j and the number of bits in each subcarrier k. It is also assumed that one requires more important frames to exactly fulfill the required rate, while the least important frame is allowed to vary slightly in order to exactly fulfill the total target rate.
• FIG. 2 models a special case of three classes of protections (Classo, Classi, and CIaSS 2 ), where Classo is the highest protected class.
• the upper graph in Fig. 2 shows the original measured S ⁇ Rs over the carrier indices, whereas the lower graph shows reordered carriers with their S ⁇ Rs after S ⁇ R sorting according to claims 4 and 7.
• the arrows added in the lower graph symbolize the movements of border lines according to claims 7 and the special realizations according to claims 10, 11.
• a horizontal movement in the ordered (lower) graph corresponds at the same time to a corresponding vertical movement in S ⁇ R direction, which would be the same in the upper graph.
• the carriers may be sorted according to their signal-to-noise ratios (" S ⁇ R sorting" ) .
• Figure 2 shows such thresholds devoting carriers to protection classes.
• S ⁇ R sorting according to claims 4 and 7 these thresholds represent thresholds at carrier numbers.
• the invention reaches the required bit load T j per class j up to some tolerance which is usually determined by the bit-load of the carriers near the borderline. In a typical embodyment of the invention, one would choose the tolerances such that the number of bits will only be decreased in the low-priority class, achieving better protection for some bits, moved to better classes by the borderline shift.
• the said S ⁇ R sorting according to Claim 4 has relations, but still differs from the so-called Tone Ordering in the ADSL (e.g., ITU-T G.992.1) and VDSL (e.g., ETSI TS 101270-2) standards patented in the United States Patent 5,596,604, Cioffi et al., Multicarrier modulation transmission system with variable delay, Jan. 21, 1997.
• Tone-Ordering the difference in protection is provided by two differently encoded channels, e.g., in ADSL a high and low latency path (slow and fast channels, respectively), with or without an interleaver, respectively.
• b k ⁇ is the bit rate for the k th carrier in the j th priority class, such that, e g , j G [O 1 . . , N 9 - I) would correspond to N 9 protection levels ⁇ 3 is fixed for a set of carriers k E M 3 .
• the number of bits, b k , ⁇ are rounded (truncated) to the nearest integer 6 fcj , where ⁇ b k,3 is the 'quantization error', and b max are the maximum allowed bits per subcarrier
• Each group is composed of a certain number of bits T 3 such that the total target bit- rate is given by
• Figure 1 depicts a possible algorithm flowchart, without including the modification according to claims 5 and 6 and specializing to a binary search of the region limits according to Claim 11 instead of the linear search according to Claim 10
• the complete algorithm is the following
• the margin recalculation in the outer loop is specified in 60 as 7o
• ne ⁇ • 2 °N
• a branching 80 leads to the final power allocation according to Claim 13 using the well-known symbol error ratio equations for given signal alphabets as, e g , given in J G Proakis, Digital Communications, McGraw Hill Higher Education Dec 1, 2000, solving them for the power for a given error ratio, which may be symbol or bit error ratios
• the power for each subcarrier will then appear as a function of the noise power at each subcarrier N fc]J , the rounded number of bits b k 3 , and the required average error probability V M ( ⁇ 3 ) for the j th class
• An ambodyment of the invention can include a further power normalization according to given power constraints
• Another embodyment may provide an outer loop, if the overall bit-error probability requirements have finally not be achieved, reducing the bit- load and reentering the algorithm according to claims 1 to 13.
• the outer loop is not shown in the flow chart in Fig. 1.
• an embody- ment of the invention may provide a branching 90 leading to a rate-enforcing step 100.
• This step is characterized in that the desired total and individual bit counts are enforced by either reducing the bit count at carrier locations where the quantization error ⁇ / tj is small or increasing the bit count at those carrier locations where the quantization error Ab ⁇ j is big.
• the invention only requires this rate-forcing step for the lowest protected class.
• the other rates are enforced by the border shifts between carrier ranges according to Claim 7 with possible search procedures according to claims 10 or 11.
• Other embodyments of the invention may choose other steps, i.e., other successive modifications of the border lines.
• the inner iterations modifying borders of carrier ranges corresponding to the different protection classes are determined by 130-170, where the boundaries are successively determined for each class.
• 180,190 are provided to flag, if no bits have been placed in the last protection class, meaning that these bits have all been better protected than required.
• the power of each subcarrier is allocated according to the quantized bit loading results. This means that the power mask will not stay perfectly constant as in the equal bit loading scheme. It will rather vary with the number of allocated bits to have a saw-tooth like shape that fluctuates by around 3 dB peak-to- peak with a discontinuity at every bit-allocation step. These fluctuations are around the desired average constant power mask. One may, of course, also decide to leave the power allocation constant and accept for some bit-error ratio variations that will be taken care of by some additional error correcting code.
• Equations (6) and (8) lead to the symbol-error probability under AWGN as a function of ⁇ ⁇ and b kj is
• Kd 3 y ⁇ .. k ⁇ K 1 (I,, !.,,) (10)
• Figure 4 shows an exemplary bit loading how it relates to the signal-to-noise ratio applying S ⁇ R sorting according to Claim 4
• Fig. 5 shows the corresponding result when inverse sorting contradicting Claim 4 would be used.
• Fig. 6 shows the spacing of S ⁇ R curves realized by the invention.
• a curve that would result from a non- UEP bit loading (Chow et al. alg.) is added as a reference.
• An example for bit-error rate curves in the case of non-stationary impulse noise added after bit-allocation is shown in Fig. 7.
• a source encoder 10 is considered as the source of data of different importance, which can, of course, also be delivered from other sources.
• a channel encoder 20 may follow, which may even be realized as an unequal error protecting encoder.
• a serial-parallel converter 30 feeds the sorting block 40 reordering the data following the SNR sorting according to claims 4 to 7.
• the bit-to-symbol mapping 50 according to Claim 17 using the bit-load defined by the UEP bit-loading method according to claims 1 to 2 is next, followed by an inverse sorting to put the carriers into the right order again.
• An IFFT 70 delivers the time-domain signal to be transmitted after parallel- to-serial conversion 80.
• the bit-to-sympol mapper according to Claim 17 requires for a corresponding symbol-to-bit demapper according to Claim 18 at the receiver using the same symbol alphabet representeded by the UEP bit-loading method as herein described.
• a reverse channel (duplex) is required to communicate either the signal-to-noise ratios or the resulting bit-allocation.
• the signal according to Claim 19 resulting from the UEP bit allcation process will follow the discrete signal alphabets resulting from the herin described UEP bit- allocation process when analyzing it with a DFT (FFT).
• FFT DFT

Landscapes

• Engineering & Computer Science (AREA)
• Signal Processing (AREA)
• Computer Networks & Wireless Communication (AREA)
• Quality & Reliability (AREA)
• Detection And Prevention Of Errors In Transmission (AREA)

Priority Applications (1)

Applications Claiming Priority (3)

Publications (1)

Family

ID=37605250

Family Applications (2)

Family Applications Before (1)

Country Status (2)

Families Citing this family (13)

Family Cites Families (5)

• 2006
  • 2006-08-30 EP EP06119771A patent/EP1895732A1/de not_active Withdrawn
• 2007
  • 2007-08-28 WO PCT/EP2007/007503 patent/WO2008025510A1/en active Application Filing
  • 2007-08-28 EP EP07801926A patent/EP2062412A1/de not_active Ceased

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events