Butterfly processor for telecommunications
View Patent ↗The present invention discloses a butterfly processor capable of performing convolutional decoding and LogMAP decoding in telecommunications systems. First and second add-compare-select modules are provided for receiving input path metrics. A branch metric calculator is also provided for receiving input data and extrinsic data. The branch metric calculator generates output branch metrics to each of the first and second add-compare-select modules. Each of the add-compare-select modules includes a log-sum correction means coupled to compare and select components. A controllable switch selectively couples outputs of the select components and the log-sum corrections means to enable either one of convolutional or LogMAP decoding.
1. Butterfly processor arrangement for use in telecommunications decoding, said arrangement comprising:
first and second add-compare-select modules, said modules each receiving first and second input path metrics supplied to said arrangement and wherein each said module comprises log-sum correction means coupled to compare and select components of said module and a controllable switch for selectively coupling outputs of said select component and said log-sum correction means to an output of said module;
a branch metric calculator, said branch metric calculator receiving input data and extrinsic data and arranged to generate first and second output branch metrics supplied to each of said add-compare-select modules, said second output branch metric being the arithmetic inverse of the first output branch metric; and
means for actuating said switch such that when said select component is output, said arrangement operates for convolutional decoding and when said log-sum correction means is output said arrangement operates for LOGMAP decoding.
2. An arrangement according to claim 1 , wherein said log-sum correction means comprises a log-sum correction table having an address input connected to a difference output of said compare component, said table outputting a log-sum value to a first input of an adder, a second input of which being coupled to the output of the select component, said adder providing an output of said log-sum correction means.
3. A butterfly processor arrangement for use in telecommunications decoding, said arrangement comprising:
first and second add-compare-select means, each said add-compare-select means for receiving first and second input path metrics supplied to said arrangement and wherein each said add-compare-select means includes log-sum correction means coupled to compare and select components of said add-compare-select means;
a controllable switch means for selectively coupling outputs of said select component and said log-sum correction means to an output of said add-compare-select means;
branch metric calculator means for receiving input data and extrinsic data and arranged to generate first and second output branch metrics supplied to each of said add-compare-select means, said second output branch metric being the arithmetic inverse of the first output branch metric; and
means for actuating said switch means such that when said select component is output, said arrangement operates for convolutional decoding and when said log-sum correction means is output said arrangement operates for LOGMAP decoding.
4. An arrangement according to claim 1 , wherein said log-sum correction means comprises a log-sum correction table means having an address input connected to a difference output of said compare component, said table means outputting a log-sum value to a first input of an adder means, a second input of which being coupled to the output of the select component, said adder means providing an output of said log-sum correction means.
5. A method for performing butterfly processing in telecommunications decoding, said method comprising the steps of:
generating first and second branch metrics from input data and extrinsic data, said second branch metric being the arithmetic inverse of said first branch metric;
performing add-compare-select operations on first and second input path metrics and said first and second branch metrics;
log-sum correcting a select operation output to produce a log-sum corrected output; and
selectively coupling said log-sum corrected output and said select operation output to produce an output;
whereby when said select operation output is output, said butterfly processing operates for convolutional decoding and when said log-sum corrected output is output, said butterfly processing operates for LOGMAP decoding.
6. The method according to claim 5 , whereby said log-sum correcting utilises a log-sum correction table, the method comprising the further steps of:
providing a difference output of said compare operation to an address input of said log-sum correction table to output a corresponding log-sum value; and
adding said log-sum value to said select operation output to provide said log-sum corrected output.