Metrics Calculations Utilizing Pre-Stored Values
AssigneesMStar Semiconductor, Inc., MStar Software R&D (Shenzhen) Ltd., MStar France SAS
InventorCyril Valadon
Life of this patent
Estimating the term…
Loading patent…
AssigneesMStar Semiconductor, Inc., MStar Software R&D (Shenzhen) Ltd., MStar France SAS
InventorCyril Valadon
Estimating the term…
A trellis processing technique for estimating a transmitted signal from a received signal acquired through a channel achieves a reduced computational structure when implementing a DDFSE-like branch metrics calculation by pre-storing computed values, which can be re-used. The technique involves storing differences between a received symbol and different products, each of which is a product of a channel coefficient corresponding to a least time delay and a different possible transmitted symbol. A summation of products is calculated, each product being a product of a possible transmitted signal symbol and a channel coefficient other than the channel coefficient corresponding to the least time delay, the summation being specific to a branch-start state in a trellis. The summation is combined with different stored differences in a calculation of branch metrics leading to different branch-end states in the trellis.
Trellis processing apparatus for calculating branch metrics in a reduced state trellis comprising a processor adapted to pre-calculate for a trellis state a sum of products of channel coefficient values and signal symbol values corresponding to at least the oldest part of the section of the channel that is not hypothesized by the state such that the sum can be re-used in calculating branch metrics for different trellis branches leading out of different states and further comprising a memory for storing the products.
A trellis processing method for calculating branch metrics in a reduced state trellis comprising precalculating for a trellis state a sum of products of channel coefficient values and signal symbol values corresponding to at least the oldest part of the section of the channel that is not hypothesized by the state such that the sum can be re-used in calculating branch metrics for different trellis branches leading out of different states and storing the products in a memory.
Dependencies are inferred from the claim text.