Partial response maximum likelihood (PRML) bit detection apparatus

A partial response maximum likelihood bit detection apparatus for deriving a bit sequence from an input information signal, comprising

• input means for receiving the input information signal;
• sampling means for sampling the input information signal at sampling instants so as to obtain sample values of the input information signal at the sampling instants;
• and
• means for calculating, determining, establishing, outputting and repeating:
• the calculating includes calculating at the sampling instant for each of a plurality of states at the sampling instant, an optimum path metric value;
• the determining includes determining for each of the plurality of states, a best predecessor state at the directly preceding sampling instant, a state at the sampling instant identifying a sequence of n subsequent bits;
• the establishing includes establishing the best path from the state at the sampling instant having the lowest optimum path metric value, back in time towards a previous sampling instant via best predecessor states, established earlier for earlier sampling instants, to establish an optimum state at the previous sampling instant;
• and
• the outputting includes outputting at least one bit of the n bits of the sequence of bits corresponding to the established optimum state at the previous sampling instant;
• the repeating includes repeating the calculating, determining, establishing and outputting for a subsequent sampling instant;
• and wherein the optimum path metric value for the state at the sampling instant is obtained by:
• comparing the optimum path metric values of all possible predecessor states at the directly preceding instant of the state at the sampling instant;
• selecting the predecessor state at the directly preceding instant having the smallest optimum path metric value as the best predecessor state;
• and
• combining the optimum path metric value of the best predecessor state at the directly preceding sampling instant and a branch metric value corresponding to the state at the sampling instant, so as to obtain the optimum path metric value for the state, the branch metric value for the state being obtained from the sample value at the sampling instant and a reference amplitude, which reference amplitude has a relationship with the state.

A method of partial response maximum likelihood bit detection comprising:

• receiving the input information signal;
• sampling, the input information signal at sampling instants so as to obtain sample values of the input information signal at the sampling instants;
• calculating at the sampling instant for each of a plurality of states at the sampling instant, an optimum path metric value,
• determining for each of the plurality of states, a best predecessor state at the directly preceding sampling instant, a state at the sampling instant identifying a sequence of n subsequent bits, where n is larger than 3, and sequences of n subsequent bits having n−1 directly successive bits of the same binary value are allocated to the same state;
• establishing the best path from the state at the sampling instant having the lowest optimum path metric value, back in time towards a previous sampling instant via best predecessor states, established earlier for earlier sampling instants, to establish an optimum state at the previous sampling instant;
• outputting at least one bit of the n bits of the sequence of bits corresponding to the established optimum state at the previous sampling instant;
• and
• repeating the calculating, determining, establishing and outputting for a subsequent sampling instant.

A method of partial response maximum likelihood bit detection comprising:

• receiving the input information signal;
• sampling, the input information signal at sampling instants so as to obtain sample values of the input information signal at the sampling instants,
• calculating at the sampling instant for each of a plurality of states at the sampling instant, an optimum path metric value;
• determining for each of the plurality of states, a best predecessor state at the directly preceding sampling instant, a state at the sampling instant identifying a sequence of n subsequent bits;
• establishing the best path from the state at the sampling instant having the lowest optimum path metric value, back in time towards a previous sampling instant via best predecessor states, established earlier for earlier sampling instants, to establish an optimum state at the previous sampling instant;
• and
• outputting at least one bit of the n bits of the sequence of bits corresponding to the established optimum state at the previous sampling instant;
• repeating the calculating, determining, establishing and outputting for a subsequent sampling instant;
• and wherein the optimum path metric value for the state at the sampling instant is obtained by:
• comparing the optimum path metric values of all possible predecessor states at the directly preceding instant of the state at the instant;
• selecting the predecessor state at the directly preceding instant having the smallest optimum path metric value as the best predecessor state;
• and
• combining the optimum path metric value of the best predecessor state at the directly preceding sampling instant and a branch metric value corresponding to the state at the instant, so as to obtain the optimum path metric value for the state, the branch metric value for the state being obtained from the sample value at the sampling instant and a reference amplitude, which reference amplitude has a relationship with the state.

A partial-response maximum-likelihood bit detection apparatus for deriving a bit sequence from an input information signal, comprising:

• an input for receiving the input information signal;
• a sampling unit, for sampling the input information signal at multiple sampling instants so as to obtain respective sample values of the input information signal at the sampling instants;
• and
• a partial response maximum likelihood calculation unit, the calculation unit deriving the bit sequence from the multiple sample values;
• the calculating unit calculates respective optimum path metric values for each of a plurality of states at the current sampling instant;
• the calculating unit determines a respective best predecessor state at the directly preceding sampling instant for each of the plurality of states, a state at the current sampling instant identifying a sequence of n subsequent bits;
• the calculating unit establishes the best path from the state at the sampling instant having the lowest optimum path metric value, back in time towards a previous sampling instant via best predecessor states established earlier for earlier sampling instants, to establish an optimum state at the previous sampling instant;
• the calculating unit outputs at least one bit of the n bits of the bit sequence, the bit corresponding to the established optimum state at the previous sampling instant;
• and wherein the calculating unit includes means for allocating sequences of n subsequent bits, having n−1 directly successive bits of the same binary value, to the same state.

A partial response maximum likelihood bit detection apparatus for deriving a digital bit sequence from an input information signal, comprising:

• an input for receiving the input information signal;
• a sampling unit for sampling the input information signal at multiple sampling instants so as to obtain respective sample values of the amplitude of the input information signal at the sampling instants;
• and
• a partial response maximum likelihood calculation unit adapted to derive the bit sequence from the multiple sample values;
• the calculating unit calculates respective optimum path metric values for each of a plurality of states at the current sampling instant;
• the calculating unit determines a respective best predecessor state at the directly preceding sampling instant for each of the plurality of states, a state at a sampling instant identifying a sequence of n subsequent bits;
• the calculating unit establishes the best path from the state at the current sampling instant back in time towards a previous sampling instant via best predecessor states established earlier for earlier sampling instants in order to establish an optimum state at the previous sampling instant, the best state having the lowest optimum path metric value;
• the calculating unit outputs at least one bit of the n bits of the bit sequence, the bit corresponding to the established optimum state at the previous sampling instant;
• and wherein the calculating unit includes means for:
• comparing the optimum path metric values of all possible predecessor states at the directly preceding instant of the state at the sampling instant;
• selecting the predecessor state at the directly preceding instant having the smallest optimum path metric value as the best predecessor state;
• and
• combining the optimum path metric value of the best predecessor state at the directly preceding sampling instant and a branch metric value corresponding to the state at the sampling instant, so as to obtain the optimum path metric value for the state, the branch metric value for the state being obtained from the sample value at the sampling instant and a reference amplitude, which reference amplitude has a relationship with the state.

A partial response maximum likelihood bit detection apparatus for deriving a bit sequence from an input information signal, comprising:

• input (120) for receiving the input information signal;
• sampling unit (122) for sampling the information signal at a constant sampling frequency fs resulting in sample amplitude values a1 at respective sampling time instants t1;
• a first portion (126) of memory for storing reference amplitude values A(sk) for each state sk of K states;
• a second portion (128) of memory for storing a repsective vector V(sk, ti) for each of the K states for each sampling instant of N sequential sampling instants, the vector for a state for a sampling instant indicates the best predecessor state for that state at the directly preceding sampling instant;
• a third portion (130) of memory for storing a respective best path metric value B−PM(sk, ti) for each of the K states at each of the N sampling instants;
• a calculation unit (124) for determining bits of the bit sequence in response to a current amplitude sample value aI at a current sampling instant tI, the determining including:
• calculating for each state sk and each other state sj, branch metric values BM[sj(tI−1),sk(tI)] for the state sk, the branch metric values for the state sk depending on the current amplitude sample value aI and the reference amplitude value A(sk) for the state;
• calculating for each state sk path metric values PM(sk,ti) for the state, each of the K path metric values for the state depending on a respective branch metric value for a state of the directly preceding sampling instant and a best path metric value stored in the third memory portion for a respective state of the directly preceding sampling instant;
• for each state, comparing the K path metric values PM(sk,ti) for the state and choosing a best path metric value B−PM(sk,tI) for the state from among the K path metric values PM(sk,ti) for the state;
• storing the best path metric values chosen for each state for the current sampling instant and storing a respective vector V(sk) for each state for the current sampling instant, the vector indicating the state of the immediately previous sample used to calculate the best path metric for the state of the current sampling instant;
• comparing the K best path metric values for the respetive states for the current sampling instant and choosing a best of the best path metric value BB−PM(tI) from among the best path metric values for the current sampling instant;
• and determining one or more bits of the bit sequence by following the vectors from the best of the best path metric value for the current sample aI back N samples to one of the best path metric values for a previous sample aI−N at a previous instant t1−N;
• and wherein N is larger than 3 and K is related to N.