Patent Grant
8473830
The present disclosure generally relates to the retrieval of data from data storage mediums such as magnetic disk drives. More particularly, the present disclosure relates to a system that detects the radial incoherence generated by the movement of a read/write head over the surface of a storage medium and decodes the data stored on a data storage medium.
A modern computing system typically comprises a central processing unit, a memory unit, and supporting hardware necessary to communicate both within the system and to the outside world. Within the computing system, communication buses may exist to transmit information and instructions between the processing unit and the memory unit. To communicate with the outside world, input and output (“I/O”) devices may be used. Common examples of input devices include the keyboard and mouse. Output devices include printers and display devices such as a monitor. In addition, storage devices such as hard disk drives, floppy disk drives, and optical disk drives may serve as both input and output devices.
In conventional hard disk drives, information in the form of analog or digital data is recorded on concentric tracks on a magnetic disk. The disk is spun at a very high speed while a read/write head moves radially over the surface of the disk in order to read information from or write information to the tracks. Information which is read out from a disk may be supplied to a device known as a Viterbi decoder to generate a reproduction of an original data signal recorded on the disk.
One type of information commonly written to and read from the tracks is servo information. Servo information may comprise control information such as address and position information to allow proper alignment of the read/write head with respect to the tracks, as well as other track identification information. Servo information is often written on a portion of the disk platter known as a servo wedge. Within a servo wedge, the servo information is broken up into multiple servo fields, each typically containing a different type of servo information. While reading such information or performing a seek operation to pinpoint a specific track, a read/write head may radially traverse multiple tracks. When the servo wedges are not properly aligned along the radial direction, “radial incoherence” (RI) may result. RI is the timing offset that occurs when a head moves between radially adjacent tracks in which the servo wedges are not properly aligned. The waveform signal read by the head may undergo a phase shift due to the RI, called the RI phase, which leads to errors in decoding the recorded servo information.
Typically, RI is detected by a RI detector with a fixed or symmetric RI detection threshold in a servo channel. However, a RI detection threshold that is set too high may fail to detect RI when it exists. Conversely, a RI detection threshold that is set too low may detect false positives when no RI exists. In addition, noise or other errors can corrupt a sequence of RI data.
Thus, there is a need for an improved system for RI detection that increases the probability of detecting RI and reduces the bit error rate (BER) of a decoded signal read from a disk.
Disclosed is a system for decoding data stored on a data storage medium. The system comprises a first input terminal receiving a data stream read from the data storage medium. The data stream comprises a plurality of data transitions. The system further comprises an input unit identifying portions of the data stream as possible transitions and selecting a subset of the possible transitions in response to a selection input. A second input unit generates estimated phases of the possible transitions and selects one of the estimated phases. A phase buffer stores a plurality of the selected estimated phases. An adder generates a first sum by adding the plurality of the selected estimated phases. The system also comprises a decision unit using the first sum to generate a flag indicating a type of radial incoherence associated with the data stream and a filter generating and supplying the selection input to the first input unit.
Further disclosed is a system for decoding data stored on a data storage medium. The system comprises an input terminal receiving a data stream read from the data storage medium, the data stream comprising a plurality of data transitions. The system further comprises an input unit identifying portions of the data stream as possible transitions and selecting a subset of the possible transitions in response to a selection input. A path-based input unit generates a first sum and a path-based threshold unit generates a second sum. The system also comprises a decision unit using the first and second sums to generate a flag indicating a type of radial incoherence associated with the data stream and a filter generating and supplying the selection input to the first input unit.
Also disclosed is a method for decoding data stored on a data storage medium. The method comprises receiving a data stream read from the data storage medium, the data stream comprising a plurality of data transitions. The method further comprises identifying portions of the data stream as possible transitions and selecting a subset of the possible transitions in response to a selection input. The method also comprises estimating phases of the possible transitions and selecting one of the estimated phases. The method further comprises generating a first sum by adding a plurality of the selected estimated phases. The method also comprises generating a plurality of flags indicating types of radial incoherence associated with the data stream and generating the selection input from the plurality of flags.
Also disclosed is a method for decoding data stored on a data storage medium. The method comprises receiving a data stream read from the data storage medium, the data stream comprising a plurality of data transitions. The method further comprises identifying portions of the data stream as possible transitions and selecting a subset of the possible transitions in response to a selection input. The method also comprises estimating phases of the possible transitions and generating a first sum from the plurality of estimated phases. The method further comprises storing a plurality of path-based threshold values and generating a second sum from the plurality of path-based threshold values. The method also comprises generating a plurality of flags indicating types of radial incoherence associated with the data stream and generating the selection input from the plurality of flags.
The accompanying drawings, which are incorporated in and constitute a part of this specification, show certain aspects of implementations consistent with the present invention and, together with the description, help explain some of the principles associated with the invention. In the drawings,
Reference will now be made in detail to the invention, examples of which are illustrated in the accompanying drawings. The implementations set forth in the following description do not represent all the implementations consistent with the claimed invention. Instead, they are merely some examples consistent with the present invention. Wherever possible, the same reference numbers will be used throughout the drawings to refer to the same or like parts.
Disk drive 35 also includes an actuator 41, which moves and controls an actuator arm 40 supporting a read/write head 39. In response to control by actuator 41 and actuator arm 40, read/write head 39 moves in a substantially radial direction across the concentric tracks 38, as shown by arrow 42.
When playing back the servo data, it is expected that read/write head 39 moves along the center of a track 51 such that the interference from adjacent tracks will be very small and thus can be ignored. In reality, however, read/write head 39 may not always move along the center of a track, resulting in influence from adjacent tracks that cannot be ignored. In considering the influence of adjacent tracks, a linear model is frequently used. The signal at the output of read/write head 39 may be modeled as:
z(t)=(1−α)*x(t)+α*y(t) (1)
where z(t) is the signal at the output of read/write head 39, x(t) is the on-track (i.e., the target track) signal, y(t) is the off-track (i.e., one of the adjacent tracks) signal, and 0≦α≦1 is the position of read/write head 39. If α=0, read/write head 39 is at the center of on-track. If α=0.5, read/write head 39 is at the middle of the two tracks, on-track and off-track. If α=1, read/write head 39 is at the center of off-track.
The signal read by read/write head 39 is in the form of a continuous-time waveform. This signal will usually pass through an analog front end (AFE), an analog to digital converter (ADC), and a digital equalizer. The signal at the output of the digital equalizer, which is the input signal to the Viterbi decoder, may be modeled as:
z′(k)=(1−α)*x′(k)+α*y′(k) (2)
where z′(k) is the signal at the output of the digital equalizer and the input signal to the Viterbi decoder and x′(k) and y′(k) are the samples at the equalizer output for signals x(t) and y(t), respectively, at the k-th sampling instant, where k is an integer. From Equation (2), it is evident that z′(k) will be different than the desired signal x′ (k) as long as y′ (k) x′ (k) and α≠0. This effect is called RI as it is due to the radial offset of the read/write head 39 from the center of the on-track.
An exemplary servo information format 50, servo field 54 contains a track-ID in GC and the GC of two radially adjacent servo fields 54 in a same servo wedge only differ by one di-bit (i.e., four coded bits), and each of the servo fields 52, 53, and 55 on a same servo wedge are the same. Thus, the waveforms of the servo fields on a servo wedge across radially adjacent tracks will have the same shape. The waveforms will differ only in initial phase. Accordingly, y(t), the off-track signal, can be expressed as:
y(t)=x(t−τ) (3)
where τ is a real value and represents the phase difference. Negative RI occurs where τ<0 and α≠0. Positive RI occurs where τ>0 and a 0.
Trajectory 62 represents the relative movement of read/write head 39 directly between tracks 51b and 51c. Because of the read/write head's position relative to tracks 51b and 51c, it senses a signal from both tracks. Though the signal that read/write head 39 picks up combines the signals from track 51b and 51c, RI does not exist because the servo fields in the servo wedge are aligned and there is no phase shift between the signals picked up from the tracks.
Trajectory 64 represents the relative movement of read/write head 39 including radial movement across tracks from track 51b to 51a. As read/write head 39 moves between tracks 51b and 51a, the signal from track 51b grows weaker as the head moves away from the center of the track and the signal from track 51a grows stronger as the head moves toward the center of track 51a. However, RI does not exist because the servo fields are aligned, thus there is no phase shift between signals from the two radially adjacent tracks. In
The relative movement of read/write head 39 is shown along different trajectories 70-74. Trajectory 72 represents the relative movement of read/write head 39 along the center of track 51e, where track 51e is the on-track. Because the read/write head is moving directly along the center of the track, the head picks up the strongest signal from track 51e. Any signals the head may sense from adjacent tracks 51d and 51f (i.e., off-track signals) are weak. RI does not exist along trajectory 72 since read/write head 39 picks up a strong signal from track 51e and negligible signals from adjacent tracks. In trajectory 72, z′(k)=x′(k) because α=0.
Trajectory 70 represents the relative movement of read/write head 39 directly between tracks 51e and 51f. Because of the read/write head's position relative to tracks 51e and 51f, it senses a signal from both tracks. Relative to the position of the servo fields on track 51e, the servo fields on track 51f are farther along the polar axis in the direction of the read/write head's relative movement. Accordingly, read/write head 39 senses a signal from track 51f with a positive phase shift relative to that of the signal it senses from track 51e. This phase shift between the two signals causes destructive interference and the combined signal picked up by read/write head 39 will have positive RI. Positive RI occurs when read/write head 39 is in the middle of two radially adjacent servo fields of a servo wedge on radially adjacent tracks. When this occurs α=0.5. Both the on-track and off-track signals have the same shape, but the off-track signal is delayed by one coded bit cycle with respect to the on-track signal. The off-track signal can be expressed as:
y(t)=x(t−T) (4)
where T is the coded bit cycle.
Trajectory 73 represents the relative movement of read/write head 39 first along the center of track 51e and then in a radial direction across tracks from track 51e to 51f. While read/write head 39 moves along the center of track 51e, the signal picked up by read/write head 39 is the on-track signal and has no RI. As read/write head 39 moves between the two tracks 51e and 51f, the signal from track 51e grows weaker as the head moves away from the center of the track and the signal from track 51f (i.e., off-track signal) grows stronger as the head moves toward the center of track 51f. Relative to the position of the servo fields on track 51e, the servo fields on track 51f are farther along the polar axis in the direction of the read/write head's relative movement. Accordingly, read/write head 39 senses a signal from track 51f with a positive phase shift relative to that of the signal it senses from track 51e. This phase shift between the two signals causes destructive interference and the combined signal picked up by read/write head 39 as it moves radially across the tracks will have positive radial incoherence.
Trajectory 71 represents the relative movement of read/write head 39 moving directly between tracks 51e and 51d. Because of the read/write head's position relative to tracks 51e and 51d, it senses a signal from both tracks. Relative to the position of the servo fields on track 51e, the servo fields on track 51f are not as far along the polar axis in the direction of the read/write head's relative movement. Accordingly, read/write head 39 senses a signal from track 51d with a negative phase shift relative to that of the signal it senses from track 51e. This phase shift between the two signals causes destructive interference and the combined signal picked up by read/write head 39 will have negative radial incoherence. Just like positive RI, negative RI occurs when read/write head 39 is in the middle of two radially adjacent servo fields of a servo wedge on radially adjacent tracks. When this occurs α=0.5. Both the on-track and off-track signals have the same shape, but the off-track signal precedes the on-track signal by one coded bit cycle. The off-track signal can be expressed as:
y(t)=x(t+T) (4)
where T is the coded bit cycle.
Trajectory 74 represents the relative movement of read/write head 39 in a radial direction across tracks from track 51e to 51d. As read/write head 39 moves between the two tracks 51e and 51d, the signal from track 51e grows weaker as the head moves away from the center of the track and the signal from track 51d grows stronger as the head moves toward the center of track 51d. Relative to the position of the servo fields on track 51e, the servo fields on track 51d are not as far along the polar axis in the direction of the read/write head's relative movement. Accordingly, read/write head 39 senses a signal from track 51d with a negative phase shift relative to that of the signal it senses from 51e. This phase shift between the two signals causes destructive interference and the combined signal picked up by read/write head 39 as it moves radially across the tracks will have negative radial incoherence.
Decoder 85 receives an input data stream at an input terminal 90. Input data stream may comprise x0[n], x1[n], x2[n], and x3[n], a user bit that corresponds to four coded bits and represents one transition. The input data stream originates as an analog signal picked up by read/write head 39 (
In decoding the input data stream, decoder 85 may employ a trellis diagram, also called a state transition diagram, to determine a bit sequence most likely to have been recorded on the storage medium out of all possible bit sequences that may have been recorded. The PR4 and MFPR4 targets use different trellis diagrams.
The PR4 target response may be expressed as:
H(D)=1+0D−D2 (5)
where H(D) is a transfer function of the PR4 target and D is an integer coefficient implementing the transfer function. The PR4 target may be represented as a vector of tap weights [2, 0, −2] where a scaling factor of 2 has been used.
Referring to
The memory length of a PR4 target is one user bit, which corresponds to four coded bits. Accordingly, each branch 111-114 can be represented by the four ideal output samples that correspond to the particular transition. The 0→0 transition shown by branch 111 may be represented as (2, 2, −2, −2). The 0→1 transition shown by branch 112 may be represented as (2, 2, 0, 0). The 1→0 transition shown by branch 113 may be represented as (−2, −2, 0, 0). The 1→1 transition shown by branch 114 may be represented as (−2, −2, 2, 2). A path through the trellis is represented by a sequence of state transitions or branches over time.
The trellis diagram 110 of
The MFPR4 target response may be expressed as:
H(D)=−1+0D+2D2+0D3−1D4 (5)
where H(D) is a transfer function of the MFPR4 target and D is an integer coefficient implementing the transfer function. The MFPR4 target may be represented as a vector of tap weights [−1, 0, 2, 0, −1].
Referring to
The memory length of a MFPR4 target is one user bit, which corresponds to four coded bits. Accordingly, each branch 121-124 can be represented by the four ideal output samples that correspond to the particular transition. The 0→0 transition shown by branch 121 may be represented as (2, 2, −2, −2). The 0→1 transition shown by branch 122 may be represented as (1, 1, 1, 1). The 1→0 transition shown by branch 123 may be represented as (−1, −1, −1, −1). The 1→1 transition shown by branch 124 may be represented as (−2, −2, 2, 2). A path through the trellis is represented by a sequence of state transitions or branches over time.
The trellis diagram 120 of
Referring again to
The Euclidean-based branch metrics calculations associated with each transition, assuming a signal with no RI, for a PR4 target may be written as:
0→0−2x0[n]−2x1[n]+2x2[n]+2x3[n]+8 (6)
1→0 2x9[n]+2x1[n]+4
0→1−2x0[n]−2x1[n]+4
1→1 2x0[n]+2x1[n]−2x2[n]−2x3[n]+8
where x0[n], x1[n], x2[n], and x3[n] are the individual data points in the input data stream at an n-th transition.
The Euclidean-based branch metrics calculations associated with each transition, assuming an input signal with negative RI, for a PR4 target may be written as:
0→0−2x0[n]+2x[n]+4 (7)
1→0 2x0[n]+x1[n]+2.5
0→1−2x0[n]−x1[n]+2.5
1→1 2x0[n]−2x2[n]+4
where x0[n], x1[n], x2[n], and x3[n] are the individual data points in the input data stream at an n-th transition.
The Euclidean-based branch metrics calculations associated with each transition, assuming an input signal with positive RI, for a PR4 target may be written as:
0→0−2x1[n]+2x3[n]+4 (8)
1→0 2x1[n]+x2[n]+2.5
0→1−2x1[n]−x2[n]+2.5
1→1 2x1[n]−2x3[n]+4
where x0[n], x1[n], x2[n], and x3[n] are the individual data points in the input data stream at an n-th transition.
The Euclidean-based branch metrics calculations associated with each transition, assuming a signal with no RI, for a MFPR4 target may be written as:
0→0−2x0[n]−2x1[n]+2x2[n]+2x3[n]+8 (9)
1→0 x0[n]+x1[n]+x2[n]+x3[n]+2
0→1−x0[n]−x1[n]−x2[n]−x3[n]+2
1→1 2x0[n]+2x1[n]−2x2[n]−2x3[n]+8
where x0[n], x1[n], x2[n], and x3[n] are the individual data points in the input data stream at an n-th transition.
The Euclidean-based branch metrics calculation associated with each transition, assuming a signal with negative RI, for a MFPR4 target may be written as:
0→0−2x0[n]+2x2[n]+4 (10)
1→0 x0[n]+x1[n]+x2[n]+1.5
0→1−x0[n]−x1[n]−x2[n]+1.5
1→1 2x0[n]−2x2[n]+4
where x0[n], x1[n], x2[n], and x3[n] are the individual data points in the input data stream at an n-th transition.
The Euclidean-based branch metrics calculations associated with each transition, assuming a signal with positive RI, for a MFPR4 target may be written as:
0→0−2x2[n]+2x[n]+4 (11)
1→0 x2[n]+x2[n]+x3[n]+1.5
0→1−x2[n]−x2[n]−x3[n]+1.5
1→1 2x2[n]−2x[n]+4
where x0[n], x1[n], x2[n], and x3[n] are the individual data points in the input data stream at an n-th transition.
Branch metric unit 88 supplies a set of branch metrics for each identified transition, influenced by positive RI, negative RI, and no RI to branch metric selector 92 within input unit 91. Branch metric selector 92 is coupled to branch metric unit 88 and decision unit 100 within RI detector 89. From decision unit 100 branch metric selector 92 receives a RI flag 101 that can be shown as RI[n-N], where n represents the n-th input transition and N is a positive integer and represents the overall pipeline latency with respect to the input data signal at input terminal 90. Based on whether RI flag 101 indicates positive RI, negative RI, or no RI, branch metric selector 92 selects a corresponding subset of transitions 93 from the transitions identified by branch metric unit 88.
Branch metric selector 92 is coupled to ACS unit 94 and supplies subset of transitions 93 to ACS unit 94, which determines possible surviving paths from the transitions in subset of transitions 93. The possible surviving paths evolve from previously determined surviving paths. The path metric value for each possible surviving path is calculated using the branch metric of each transition in subset of transitions 93 and the path metric values of previously determined surviving paths, where the path metric values for the previously determined surviving paths are stored in PM0107 and PM1108 in output unit 95. Comparing the path metric values for each possible surviving path merging at a same state, ACS unit 94 determines a surviving path associated with each trellis state. The path metric for the surviving paths are stored in PM0107 and PM1108 in output unit 95. This is equivalent to extending each previously determined surviving path by adding one new branch to it, where the newly added branches are selected from subset of transitions 93. ACS unit 94 then compares the path metrics for the surviving paths to determine a best surviving path. ACS unit 94 selects a most recent transition in the best surviving path, a selected transition 96, which is one of the newly added branches. Selected transition 96 is supplied to output unit 95 and also to phase selector 98 within second input unit 97.
ACS unit 94 is coupled to output unit 95 and supplies the path metric value for the surviving path associated with each trellis state, the newly added branches, the best surviving path, and selected transition 96 to output unit 95. PM0107 in output unit 95 stores the path metric value for the surviving path associated with state 0. PM1108 in output unit 95 stores the path metric value for the surviving path associated with state 1. Access and operation control unit (control unit) 109 may extend the previously determined surviving paths stored in path buffers 125 and 126 by first shuffling them based on the path metric value for the surviving path associated with each trellis state and then adding, at the end of the previously determined surviving paths, the newly determined branches. Control unit 109 keeps track of the best surviving path.
Output unit 95 may also compare a plurality of subset of transitions 93 and a plurality of selected transitions 96, the best surviving path, to generate a decoded data signal 104. For instance, using the best surviving path and its path metric value, output unit 95 may follow every transition in the best surviving path from a current time backwards to a fixed trace depth to determine the correctness of the best surviving path calculated by ACS unit 94. Output unit 95 then constructs a corrected best surviving path from which it outputs a decoded data signal 104.
Input terminal 90 also supplies second input unit 97, within RI detector 86, with the input data stream. Phase estimator 87 within second input unit 97 generates an estimated phase from the input data stream for each possible transition within the input data stream. For example, given the two states 0 and 1, phase estimator 87 generates phase estimates for the following four transitions: 0→0; 0→1; 1→0; and 1→1. Like branch metric unit 88, phase estimator 87 may use a set of well-known predetermined and preprogrammed calculations or it may use a variety of calculations adaptive to several conditions to generate the estimated phase for each transition.
The RI phase associated with each transition for a PR4 target may be estimated by using:
0→0(−x0[n]+x1[n]+x2[n]−x3[n])/wt12 (12)
1→0(x0[n]−x1[n])/2
0→1(x1[n]−x0[n])/2
1→1(x0[n]−x1[n]+x3[n]−x2[n])/wt12
where x0[n], x1[n], x2[n], and x3[n] are the individual data points in the input data stream at an n-th transition and wt12 is a scaling factor used to balance the estimated phase and its variance for different transitions.
The RI phase associated with each transition for a MFPR4 target may be estimated by using:
0→0(−x0[n]+x1[n]+x2[n]−x3[n])/wt12 (13)
1→0 x9[n]−x3[n]
0→1 x0[n]+x3[n]
1→1(x0[n]−x1[n]−x2[n]+x3[n])/wt12
where x0[n], x1[n], x2[n], and x3[n] are the individual data points in the input data stream at an n-th transition and wt12 is a scaling factor used to balance the estimated phase and its variance for different transitions.
Phase estimator 87 supplies the generated estimated phases to phase selector 98 within second input unit 97. Phase selector 98 also receives selected transition 96 from ACS unit 94 or output unit 95. In response to selected transition 96, phase selector 98 selects an estimated phase 99.
Phase selector 98 supplies selected estimated phase 99 to phase buffer 84. Phase buffer 84 is coupled to phase selector 98 and phase adder 127. Phase buffer 84 receives and stores the M most recent values of selected estimated phase 99, where M is a positive integer. The larger the value of M, the more any noise in the input data signal will be compressed. However, the larger the value of M, the slower the response of decoder 85 and RI detector 86 to a RI type change. Phase buffer 84 sends the M most recent values of selected estimated phase 99 to phase adder 127.
Phase adder 127 calculates a moving sum ph_sum[n] 128 of the M most recent values of selected estimated phase 99 from the phase buffer 84. Decision unit 100 receives ph_sum[n] 128 from phase adder 127. Decision unit 100 may use a fixed threshold value, a set of fixed threshold values, or other ways of obtaining a detection threshold value consistent with the present invention to determine whether RI exists and if so, whether it is positive or negative. For example, decision unit 100 may compare the value of ph_sum[n] 128 against two present threshold values, −THR and +THR, to determine whether there exists positive RI, negative RI, or no RI. Based on the comparison, decision unit 100 generates RI flag 101, which indicates positive RI, negative RI, or no RI. Decision unit 100 supplies RI flag 101 to branch metric selector 92 within input unit 91. Object 102 between decision unit 100 and branch metric selector 92 is merely illustrative of the overall pipeline latency N with respect to the input data signal at input terminal 90.
Decoder 80 receives an input data stream conditioned in accordance with a PR4 or MFPR4 target, as previously described, at an input terminal 90. In decoding the input data stream, decoder 80 may employ trellis diagrams, as shown in
Referring again to
Branch metric unit 88 supplies a set of branch metrics for each identified transition, influenced by positive RI, negative RI, and no RI to branch metric selector 92 within input unit 91. Branch metric selector 92 is coupled to branch metric unit 88 and filter 105 within RI detector 89. From filter 105, branch metric selector 92 receives a RI type signal 106. Based on whether RI type signal 106 indicates positive RI, negative RI, or no RI, branch metric selector 92 selects a corresponding subset of transitions 93 from the transitions identified by branch metric unit 88.
Branch metric selector 92 is coupled to ACS unit 94 and supplies subset of transitions 93 to ACS unit 94, which determines possible surviving paths from the transition in subset of transitions 93. The possible surviving paths evolve from previously determined surviving paths. The path metric value for each possible surviving path is calculated using the branch metric of each transition in subset of transitions 93 and the path metric values of previously determined surviving paths, where the path metric values for the previously determined surviving paths are stored in output unit 95. Comparing the path metric values for each possible surviving path merging at a same state, ACS unit 94 determines a surviving path associated with each trellis state. The path metric for the surviving paths are stored in output unit 95. This is equivalent to extending each previously determined surviving path by adding one new branch to it, where the newly added branches are selected from subset of transitions 93. ACS unit 94 then compares the path metrics for the surviving paths to determine a best surviving path. ACS unit 94 selects a most recent transition in the best surviving path, a selected transition 96, which is one of the newly added branches. Selected transition 96 is supplied to output unit 95 and also to phase selector 98 within second input unit 97.
ACS unit 94 is coupled to output unit 95 and supplies the path metric value for the surviving path associated with each trellis state, the newly added branches, the best surviving path, and selected transition 96 to output unit 95. As described above, output unit 95 may extend the previously determined surviving paths and may also generate a decoded data signal 104.
Input terminal 90 also supplies second input unit 97, within RI detector 89, with the input data stream. Phase estimator 87 within second input unit 97 generates an estimated phase from the input data stream for each possible transition within the input data stream. For example, given the two states 0 and 1, phase estimator 87 generates phase estimates for the following four transitions: 0→0; 0→1; 1→0; and 1→1. Like branch metric unit 88, phase estimator 87 may use a set of well-known predetermined and preprogrammed calculations or it may use a variety of calculations adaptive to several conditions to generate the estimated phase for each transition. Phase estimator 87 may also estimate the phase associated with each transition using Equations (12) and (13), as described above.
Phase estimator 87 supplies the generated estimated phases to phase selector 98 within second input unit 97. Phase selector 98 also receives selected transition 96 from ACS unit 94. In response to selected transition 96, phase selector 98 selects an estimated phase 99.
Phase selector 98 supplies selected estimated phase 99 to phase buffer 84. Phase buffer 84 is coupled to phase selector 98 and phase adder 127. Phase buffer 84 receives and stores the M most recent values of selected estimated phase 99, where M is a positive integer. The larger the value of M, the more any noise in the input data signal will be compressed. However, the larger the value of M, the slower the response of decoder 80 and RI detector 89 to a RI type change. Phase buffer 84 sends the M most recent values of selected estimated phase 99 to phase adder 127.
Phase adder 127 calculates a moving sum ph_sum[n] 128 of the M most recent values of selected estimated phase 99 from the phase buffer 84. Decision unit 100 receives ph_sum[n] 128 from phase adder 127. Decision unit 100 may use a fixed threshold value, a set of fixed threshold values, or other ways of obtaining a detection threshold value consistent with certain aspects related to the present invention to determine whether RI exists and if so, whether it is positive or negative. For example, decision unit 100 may compare the value of ph_sum[n] 128 against two present threshold values, −THR and +THR, to determine whether there exists positive RI, negative RI, or no RI. Based on the comparison, decision unit 100 generates RI flag 101, which indicates positive RI, negative RI, or no RI. Decision unit 100 supplies RI flag 101 to filter 105. Object 102 between decision unit 100 and filter 105 is merely illustrative of the overall pipeline latency N with respect to the input data signal at input terminal 90.
Decision unit 100 supplies RI flag 101 to filter 105. Filter 105 is coupled between decision unit 100 and branch metric selector 92 to filter out noise or other errors which may distort or affect the RI flag. Assuming noise and other errors do not occur in burst, filter 105 corrects the error. For instance, RI flags within a sequence of RI flags stored in filter 105 or a separate buffer (not shown) may indicate positive RI at a series of discrete time instants, indicate no RI at another time instant, and then again indicate positive RI at a series of discrete time instants (ex. {1, 1, 1, 1, 1, 0, 1, 1, 1, 1}) due to error, noise, or other factors causing distortion of the RI flags. Filter 105 detects and corrects for such noise and error, generating a RI type signal 106, which may have values of 1, 0, and −1. In the above example, RI type signal 106 would be “1.” Filter 105 may be a non-linear filter such as a median filter. A separate buffer (not shown) for storing the sequence of RI flags may also be used by filter 105. In the present embodiment, filter 105 receives RI flags and stores a sequence of RI flags, compares the flags in the sequence, and generates a RI type signal 106 based upon the comparison. RI type signal 106 is then supplied to the branch metric selector 92 within input unit 91.
Decoder 135 receives an input data stream conditioned in accordance with a PR4 or MFPR4 target, as previously described, at an input terminal 90. In decoding the input data stream, decoder 135 may employ trellis diagrams, as shown in
Referring again to
Branch metric unit 88 supplies a set of branch metrics for each identified transition, influenced by positive RI, negative RI, and no RI to branch metric selector 92 within input unit 91. Branch metric selector 92 is coupled to branch metric unit 88 and decision unit 133 within RI detector 135. From decision unit 133, branch metric selector 92 receives a RI flag 101. Based on whether RI flag 101 indicates positive RI, negative RI, or no RI, branch metric selector 92 selects a corresponding subset of transitions 93 from the transitions identified by branch metric unit 88.
Branch metric selector 92 is coupled to ACS unit 94 and supplies subset of transitions 93 to ACS unit 94, which determines possible surviving paths from the transition in subset of transitions 93. The possible surviving paths evolve from previously determined surviving paths. The path metric value for each possible surviving path is calculated using the branch metric of each transition in subset of transitions 93 and the path metric values of previously determined surviving paths, where the path metric values for the previously determined surviving paths are stored in output unit 95. Comparing the path metric values for each possible surviving path merging at a same state, ACS unit 94 determines a surviving path associated with each trellis state. The path metric for the surviving paths are stored in output unit 95. This is equivalent to extending each previously determined surviving path by adding one new branch to it, where the newly added branches are selected from subset of transitions 93. ACS unit 94 then compares the path metrics for the surviving paths to determine a best surviving path. ACS unit 94 selects a most recent transition in the best surviving path, a selected transition 96, which is one of the newly added branches. Selected transition 96 is supplied to output unit 95 (not shown) and also to phase selector 98 within second input unit 97.
ACS unit 94 is coupled to output unit 95 and supplies the path metric value for the surviving path associated with each trellis state, the newly added branches, the best surviving path, and selected transition 96 to output unit 95. As described above, output unit 95 may extend the previously determined surviving paths and may also generate a decoded data signal 104.
Input terminal 90 also supplies second input unit 97, within RI detector 130, with the input data stream. Phase estimator 87 within second input unit 97 generates an estimated phase from the input data stream for each possible transition within the input data stream. For example, given the two states 0 and 1, phase estimator 87 generates phase estimates for the following four transitions: 0→0; 0→1; 1→0; and 1→1. Like branch metric unit 88, phase estimator 87 may use a set of well-known predetermined and preprogrammed calculations or it may use a variety of calculations adaptive to several conditions to generate the estimated phase for each transition. Phase estimator 87 may also estimate the phase associated with each transition using Equations (12) and (13), as described above.
Phase estimator 87 supplies the generated estimated phases to phase selector 98 within second input unit 97. Phase selector 98 also receives selected transition 96 from ACS unit 94. In response to selected transition 96, phase selector 98 selects an estimated phase 99.
Phase selector 98 supplies selected estimated phase 99 to phase buffer 84. Phase buffer 84 is coupled to phase selector 98 and phase adder 127. Phase buffer 84 receives and stores the M most recent values of selected estimated phase 99, where M is a positive integer. The larger the value of M, the more any noise in the input data signal will be compressed. However, the larger the value of M, the slower the response of decoder 135 and RI detector 130 to a RI type change. Phase buffer 84 sends the M most recent values of selected estimated phase 99 to phase adder 127.
Phase adder 127 calculates a moving sum ph_sum[n] 128 of the M most recent values of selected estimated phase 99 from the phase buffer 84. Decision unit 133 receives ph_sum[n] 128 from phase adder 127. Decision unit 133 is coupled to phase adder 127, adaptive threshold unit 131, and branch metric selector 92. Decision unit 133 also receives an adaptive threshold value 132 from adaptive threshold unit 131. Ph sum[n] 128 is compared against adaptive threshold value 132 to determine whether there exists positive RI, negative RI, or no RI. Based on the comparison, decision unit 133 generates RI flag 101, which indicates positive RI, negative RI, or no RI. Decision unit 133 supplies RI flag 101 to branch metric selector 92 and adaptive threshold unit 131. Object 102 between decision unit 133 and branch metric selector 92 and adaptive threshold unit 131 is merely illustrative of the overall pipeline latency N with respect to the input data signal at input terminal 90.
Adaptive threshold unit 131 is coupled to decision unit 133 and supplies adaptive threshold value 132 to decision unit 133. Adaptive threshold unit 131 adjusts adaptive threshold value 132 as information is gained about the RI in the system from RI flag 101. For instance, when positive or negative RI exists, a conventional threshold value may fail to detect when RI exists if it is too high. And when no RI exists, conventional threshold value may detect false positives if it is too low. Thus, adaptive threshold unit 131 may output a pre-determined adaptive threshold value 132 when no information exists about RI in the system or RI flag 101 is empty. When RI flag 101 indicates no RI, adaptive threshold unit 131 may increase adaptive threshold value 132. To achieve this result, adaptive threshold unit 131 may multiply adaptive threshold value 132 by λH, where λH is a parameter of adaptive threshold unit 131. When RI flag 101 indicates positive or negative RI, adaptive threshold unit 131 may decrease adaptive threshold value 132. To achieve this result, adaptive threshold unit 131 may multiply adaptive threshold value 132 by λL, where λL is a parameter of adaptive threshold unit 131. In the present embodiment, adaptive threshold unit 131 stores an adaptive threshold value 132, receives RI flag 101, compares RI flag 101 with the adaptive threshold value 132, adjusts the adaptive threshold value based upon the comparison, supplies the adjusted adaptive threshold value to decision unit 133, and stores the adjusted adaptive threshold value as the adaptive threshold value.
Decoder 160 receives an input data stream conditioned in accordance with a PR4 or MFPR4 target, as previously described, at an input terminal 90. In decoding the input data stream, decoder 160 may employ trellis diagrams, as shown in
Referring again to
Branch metric unit 88 supplies a set of branch metrics for each identified transition, influenced by positive RI, negative RI, and no RI to branch metric selector 92 within input unit 91. Branch metric selector 92 is coupled to branch metric unit 88 and decision unit 151 within RI detector 140. From decision unit 151, branch metric selector 92 receives RI flag 101. Based on whether RI flag 101 indicates positive RI, negative RI, or no RI, branch metric selector 92 selects a corresponding subset of transitions 93 from the transitions identified by branch metric unit 88.
Branch metric selector 92 is coupled to ACS unit 94 and supplies subset of transitions 93 to ACS unit 94, which determines possible surviving paths from the transition in subset of transitions 93. The possible surviving paths evolve from previously determined surviving paths. The path metric value for each possible surviving path is calculated using the branch metric of each transition in subset of transitions 93 and the path metric values of previously determined surviving paths, where the path metric values for the previously determined surviving paths are stored in output unit 95. Comparing the path metric values for each possible surviving path merging at a same state, ACS unit 94 determines a surviving path associated with each trellis state. The path metric for the surviving paths are stored in output unit 95. This is equivalent to extending each previously determined surviving path by adding one new branch to it, where the newly added branches are selected from subset of transitions 93. ACS unit 94 then compares the path metrics for the surviving paths to determine a best surviving path. ACS unit 94 selects a most recent transition in the best surviving path, a selected transition 96, which is one of the newly added branches. Selected transition 96 is supplied to output unit 95 (not shown) and also to phase selector 98 within second input unit 97.
ACS unit 94 is coupled to output unit 95 and supplies the path metric value for the surviving path associated with each trellis state, the newly added branches, the best surviving path, and selected transition 96 to output unit 95. As described above, output unit 95 may extend the previously determined surviving paths and may also generate a decoded data signal 104.
Input terminal 90 also supplies second input unit 97, within RI detector 140, with the input data stream. Phase estimator 87 within second input unit 97 generates an estimated phase from the input data stream for each possible transition within the input data stream. For example, given the two states 0 and 1, phase estimator 87 generates phase estimates for the following four transitions: 0→0; 0→1; 1→0; and 1→1. Like branch metric unit 88, phase estimator 87 may use a set of well-known predetermined and preprogrammed calculations or it may use a variety of calculations adaptive to several conditions to generate the estimated phase for each transition. Phase estimator 87 may also estimate the phase associated with each transition using Equations (12) and (13), as described above.
Phase estimator 87 supplies the generated estimated phases to phase selector 98 within second input unit 97. Phase selector 98 also receives selected transition 96 from ACS unit 94. In response to selected transition 96, phase selector 98 selects an estimated phase 99.
Phase selector 98 supplies selected estimated phase 99 to phase buffer 84. Phase buffer 84 receives and stores the M most recent values of selected estimated phase 99, where M is a positive integer. The larger the value of M, the more any noise in the input data signal will be compressed. However, the larger the value of M, the slower the response of decoder 160 and RI detector 140 to a RI type change. Phase buffer 84 sends the M most recent values of selected estimated phase 99 to phase adder 127. Phase adder 127 calculates a moving sum ph_sum[n] 128 of the M most recent values of selected estimated phase 99 from the phase buffer 84.
Concurrently, threshold table 142 receives selected transition 96 from ACS unit 94. Threshold table 142 contains transition-based threshold values for all possible transition. The transition-based threshold value for each transition is one half of the average value of the absolute values of the ideal RI phases for that transition under both negative and positive RI. Threshold table 142 may hold different values for the transition-based threshold under a PR4 target and under a MFPR4 target. Threshold buffer 143 within transition-based threshold unit 141 receives and stores the M most recent transition-based threshold values corresponding to selected transitions 96 from threshold table 142. The M most recent transition-based threshold values corresponding to selected transitions 96 are sent to threshold adder 144. Threshold adder 144 calculates a moving sum Thr_sum[n] 145 of the M most recent transition-based threshold values corresponding to selected transitions 96.
Decision unit 151 is coupled to phase adder 127 and threshold adder 144 and receives ph_sum[n] 128 and Thr_sum[n] 145. Decision unit 151 compares ph_sum[n] 128 against −Thr_sum[n] 145 and +Thr_sum[n] 145 to determine whether there exists positive RI, negative RI, or no RI. Based on the comparison, decision unit 151 generates RI flag 101, which indicates positive RI, negative RI, or no RI. Decision unit 151 supplies RI flag 101 to branch metric selector 92. Object 102 between decision unit 151 and branch metric selector 92 is merely illustrative of the overall pipeline latency N with respect to the input data signal at input terminal 90.
Decoder 180 receives an input data stream conditioned in accordance with a PR4 or MFPR4 target, as previously described, at an input terminal 90. In decoding the input data stream, decoder 180 may employ trellis diagrams, as shown in
Referring again to
Branch metric unit 88 supplies a set of branch metrics for each identified transition, influenced by positive RI, negative RI, and no RI to branch metric selector 92 within input unit 91. Branch metric selector 92 is coupled to branch metric unit 88 and decision unit 184 within RI detector 181. From decision unit 184, branch metric selector 92 receives RI flag 101. Based on whether RI flag 101 indicates positive RI, negative RI, or no RI, branch metric selector 92 selects a corresponding subset of transitions 93 from the transitions identified by branch metric unit 88.
Branch metric selector 92 is coupled to ACS unit 94 and supplies subset of transitions 93 to ACS unit 94, which determines possible surviving paths from the transitions in subset of transitions 93. The possible surviving paths evolve from previously determined surviving paths. The path metric value for each possible surviving path is calculated using the branch metric of each transition in subset of transitions 93 and the path metric values of previously determined surviving paths, where the path metric values for the previously determined surviving paths are stored in PM0107 and PM1108 in output unit 95. Comparing the path metric values for each possible surviving path merging at a same state, ACS unit 94 determines a surviving path associated with each trellis state. The path metric for the surviving paths are stored in PM0107 and PM1108 in output unit 95. This is equivalent to extending each previously determined surviving path by adding one new branch to it, where the newly added branches are selected from subset of transitions 93. ACS unit 94 then compares the path metrics for the surviving paths to determine a best surviving path. ACS unit 94 selects a most recent transition in the best surviving path, a selected transition 96, which is one of the newly added branches. Selected transition 96 is supplied to output unit 95.
ACS unit 94 is coupled to output unit 95 and supplies the path metric value for the surviving path associated with each trellis state, the newly added branches, the best surviving path, and selected transition 96 to output unit 95. PM0107 in output unit 95 stores the path metric value for the surviving path associated with state 0. PM1108 in output unit 95 stores the path metric value for the surviving path associated with state 1. Access and operation control unit (control unit) 109 may extend the previously determined surviving paths stored in path buffers 125 and 126 by first shuffling them based on the path metric value for the surviving path associated with each trellis state and then adding, at the end of the previously determined surviving paths, the newly determined branches. Control unit 109 keeps track of the best surviving path.
Output unit 95 may also compare a plurality of subset of transitions 93 and a plurality of selected transitions 96, the best surviving path, to generate a decoded data signal 104. For instance, using the best surviving path and its path metric value, output unit 95 may follow every transition in the best surviving path from a current time backwards to a fixed trace depth to determine the correctness of the best surviving path calculated by ACS unit 94. Output unit 95 then constructs a corrected best surviving path from which it outputs a decoded data signal 104. Output unit 95 is coupled to threshold calculation unit 189 within path-based threshold unit 182 and phase calculation unit 187 within path-based input unit 183 via link 190. Output unit 95 may send best surviving path information to threshold calculation unit 182 and phase calculation unit 187 via link 190.
Input terminal 90 also supplies path-based input unit 183, within RI detector 181, with the input data stream. Phase estimator 87 within path-based input unit 183 generates an estimated phase from the input data stream for each possible transition within the input data stream. For example, given the two states 0 and 1, phase estimator 87 generates phase estimates for the following four transitions: 0→0; 0→1; 1→0; and 1→1. Like branch metric unit 88, phase estimator 87 may use a set of well-known predetermined and preprogrammed calculations or it may use a variety of calculations adaptive to several conditions to generate the estimated phase for each transition. Phase estimator 87 may also estimate the phase associated with each transition using Equations (12) and (13), as described above.
The phase estimates generated by phase estimator 87 are sent to path-based phase buffer 186. Path-based phase buffer 186 may include a dedicated phase buffer for each transition, 0→0, 0→1, 1→0, and 1→1. These dedicated phase buffers hold the most recent phase estimates generated by phase estimator 87 for each transition and may be first-in-first-out (FIFO) buffers. After the phase estimates generated by phase estimator 87 are sent to path-based phase buffer 186, phase calculation unit 187 accesses output unit 95 via link 190 to retrieve information on a plurality of transitions in the best surviving path. Using the information on the plurality of transitions in the best surviving path, phase calculation unit 187 selects a corresponding estimated phase for each transition in the plurality of transitions from the path-based phase buffer 186 and adds them together. The result, a moving sum summed phase 191, is sent to decision unit 184.
Concurrently, path-based threshold unit 182 also receives information on a plurality of transitions in the best surviving path. Path-based threshold unit 182 includes threshold table 188 which contains transition-based threshold values for all possible transitions with positive RI and negative RI. Accordingly, each transition has a transition-based threshold value associated with positive RI and a transition-based threshold value associated with negative RI. The transition-based threshold value for each transition is one half of the theoretical RI phase for that transition under either positive or negative RI. Threshold table 188 may hold different values for the transition-based threshold under a PR4 target and under a MFPR4 target. Alternatively, the transition-based threshold value may be calculated not just using a n-th user bit and a (n−1)-th user bit (i.e., one transition), but may be calculated using the n-th user bit, the (n−1)-th user bit, and a (n−2)-th user bit or a (n+1)-th user bit. Threshold table 188 is coupled to threshold calculation unit 189 to which it sends the transition-based threshold values corresponding to the plurality of transitions in the best surviving path.
Threshold calculation unit 189 is coupled between path-based threshold table 188 and decision unit 184 and receives transition-based threshold values corresponding to the plurality of transitions in the best surviving path. Threshold calculation unit 189 separately sums the transition-based threshold values associated with positive RI and the transition-based threshold values associated with negative RI to generate two moving sum summed thresholds 192 that it sends to decision unit 184.
Decision unit 184 is coupled to threshold calculation unit 189, phase calculation unit 187, and branch metric selector 92 within input unit 91. Decision unit 184 receives summed thresholds 192 from threshold calculation unit 189 and summed phase 191 from phase calculation unit 187. It compares summed phase 191 against summed thresholds 192 to determine whether there exists positive RI, negative RI, or no RI. Based on the comparison, decision unit 184 generates RI flag 101, which indicates positive RI, negative RI, or no RI. Decision unit 184 supplies RI flag 101 to branch metric selector 92 within input unit 91. There may exist an overall pipeline latency between RI flag 101 and the input data signal at input terminal 90.
It will be understood by one skilled in the art that filter 105, adaptive threshold unit 131, transition-based threshold unit 141, and path-based threshold unit 182 with path-based input unit 183 may be used separately or in combination. For example, filter 105 may be used alone, with adaptive threshold unit 131, with transition-based threshold unit 141, with path-based threshold unit 182 with path-based input unit 183, or with both adaptive threshold unit 131 and transition-based threshold unit 141. Other combinations are possible, for specific applications.
The systems and methods disclosed herein are not inherently related to any particular computer or other apparatus, and may be implemented by a suitable combination of hardware, software, and/or firmware. Software implementations may include one or more computer programs. A computer program is a set of instructions readable and executable by a processor and can be written in any form of programming language, including compiled or interpreted languages, and it can be deployed in any form, including as a stand alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program can be deployed to be executed on one computer or on multiple computers at one site or distributed across multiple sites and interconnected by a communication network. Software may also be implemented as a computer program product, i.e., one or more computer programs tangibly embodied in an information carrier, e.g., in a machine readable storage device or in a propagated signal, for execution by, or to control the operation of, data processing apparatus, e.g., a programmable processor, a computer, or multiple computers.
The foregoing description is intended to illustrate but not to limit the scope of the invention, which is defined by the scope of the appended claims. Other embodiments are within the scope of the following claims.
This application is a continuation of U.S. patent application Ser. No. 13/282,910, filed Oct. 27, 2011 now U.S. Pat. No. 8,261,172, which is a continuation of U.S. patent application Ser. No. 12/015,913, filed Jan. 17, 2008 (now U.S. Pat. No. 8,051,365), which claims the benefit of the filing date of U.S. Provisional Application No. 60/885,297 filed Jan. 17, 2007 (now expired), all of which are incorporated by reference herein in their entireties.
| Number | Name | Date | Kind |
|---|---|---|---|
| 7082005 | Annampedu et al. | Jul 2006 | B2 |
| 7349167 | Erden et al. | Mar 2008 | B2 |
| 7362536 | Liu et al. | Apr 2008 | B1 |
| 20030099052 | Annampedu et al. | May 2003 | A1 |
| 20050264922 | Erden et al. | Dec 2005 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 60885297 | Jan 2007 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 13282910 | Oct 2011 | US |
| Child | 13601404 | US | |
| Parent | 12015913 | Jan 2008 | US |
| Child | 13282910 | US |
