Efficient maximum-likelihood decoder for the golay (24,12) code
AssigneeUnisys Corporation
InventorsCraig K. Rushforth, Ayyoob D. Abbaszadeh
Life of this patent
Estimating the term…
Abstract
A novel maximum likeihood decoder for the Golay (24,12) code is provided. Instead of decoding the received dimensional vector (X) is a systolic array, the vector is mapped into a (24,5) subcode with an index 128 which is easily decoded. The decoder employs a plurality of precomputational circuits cooperating with a similar plurality of inner product circuits which compute the 128 inner product values and select the maximum inner product value from each inner product circuit. The maximum inner product value from each circuit is loaded into a select logic circuit for selecting the maximum inner product value and its corresponding code word closest to the received vector (X).
Claims18
An efficient maximum likelihood decoder for the Golay (24,12) code comprising:
- precomputational circuit means for receiving a dimensional vector (X) to be decoded,
- register means comprising a plurality of sets of storage registers for dividing said dimensional vector information into sets of information,
- bit serial adder means coupled to said individual sets of said registers for producing precomputed SG.sub.K values representative of said sets of information,
- random access (RAM) storage means for storing said SG.sub.K values,
- inner product circuit means comprising a plurality of processing elements,
- each of said processing elements being coupled to a part of said RAM storage means representative of sets of said SG.sub.K values,
- means for addressing and sequentially reading said precomputed SG.sub.K values out of said random access storage means into said inner product circuit means,
- means for comparing said SG.sub.K values and for determining a minimum SG.sub.K values and the subcode values for the SG.sub.K values,
- means for determining the sign of the SG.sub.K subcode values,
- means for calculating and selecting the largest inner product values for said subcode values,
- select means for determining the single subcode value whose inner product with the received dimensional vector (X) is the largest, and
- means for presenting said subcode value as information bits associated with said subcode value to a utilization device.
Dependencies are inferred from the claim text.