Systems and methods for scaling parameters
View Patent ↗The disclosed technology provides a system and a method for scaling parameters in a real-time computation system. A real-time computation system can receive a plurality of input data. The real-time computation system can determine a scaling factor based on an input data, and the scaling factor can be used in connection with a subsequent input data. When a new input data arrives, the real-time computation system can, in parallel, determine a new scaling factor based on the new input data and compute output values based on the existing scaling factor and the new input data. In one aspect of the invention, the real-time computation system can be a statistical signal processing system, which can compute output values, at least in part, using look-up tables.
1. A method for scaling values in a real-time computation system, the method comprising:
receiving a first data comprising a first plurality of estimates of noise variance values;
determining, by a scaling factor computation circuit, a first scaling factor based on a smallest one of the first plurality of estimates of noise variance values;
receiving a second data after receiving the first data, wherein the second data comprises a second plurality of estimates of noise variance value; and
performing in parallel:
computing, by a computation circuit, output data based on the first scaling factor and the second data using a ratio of (i) the smallest one of the first plurality of estimates of noise variance values to (ii) one of the second plurality of estimates of noise variance values, wherein computing the output data comprises computing, in real time, parameters for the second data based on the first scaling factor associated with the first data, and
determining a second scaling factor based on the second data.
2. The method of claim 1 , further comprising:
receiving a third data after receiving the second data; and
performing in parallel:
computing output data based on the second scaling factor and the third data, and
determining a third scaling factor based on the third data.
3. The method of claim 1 , wherein:
the real-time computation system performs non-linear Viterbi detection;
the estimates of noise variance values correspond to different filter values in the Viterbi detection; and
the output values are non-linear Viterbi detection parameters.
4. The method of claim 1 , further comprising:
computing a plurality of values based on the first scaling factor and a plurality of predetermined values; and
storing the plurality of values in a memory.
5. The method of claim 4 , wherein computing output data based on the first scaling factor and the second data comprises:
computing a memory address based on the second data; and
accessing the memory based on the memory address to access one of the plurality of values.
6. The method of claim 5 , further comprising:
computing a plurality of second values based on the second scaling factor and the plurality of predetermined values; and
storing the plurality of second values in the memory.
7. The method of claim 6 , further comprising:
receiving a third data after receiving the second data;
computing a second memory address based on the third data; and
accessing the memory based on the second memory address to access one of the plurality of second values.
8. The method of claim 7 , wherein the plurality of values and the plurality of second values are quotients.
9. A real-time computation system comprising:
an input connection that receives a first data comprising a first plurality of estimates of noise variance values and a second data after receiving the first data, wherein the second data comprises a second plurality of estimates of noise variance value;
a scaling factor memory that stores a first scaling factor;
a scaling factor computation circuit in communication with the scaling factor memory, wherein the scaling factor computation circuit determines the first scaling factor based on a smallest one of the first plurality of estimates of noise variance values; and
a computation circuit in communication with the scaling factor memory, wherein the computation circuit computes output data based on the second data and the first scaling factor in the scaling factor memory using a ratio of (i) the smallest one of the first plurality of estimates of noise variance values to (ii) one of the second plurality of estimates of noise variance values,
wherein the computation circuit computes the output data by computing, in real time, parameters for the second data based on the first scaling factor associated with the first data, and
wherein the scaling factor computation circuit and the computation circuit operate in parallel.
10. The system of claim 9 , wherein the scaling factor computation circuit comprises selection circuitry that selects one of the first plurality of estimates of noise variance values in the data.
11. The system of claim 10 , wherein the selection circuitry comprises circuitry that selects the smallest one of the first plurality of estimates of noise variance values.
12. The system of claim 11 , wherein:
the real-time computation system is a non-linear Viterbi detection system;
the estimates of noise variance values correspond to different filter values in the Viterbi detection; and
the output values are non-linear Viterbi detection parameters.
13. The system of claim 9 , further comprising:
a memory initialization circuit that computes a plurality of values based on the scaling factor and a plurality of predetermined values; and
a second memory comprising the plurality of values.
14. The system of claim 13 , wherein the computation circuit is in communication with the second memory, the computation circuit comprising:
circuitry that computes a memory address based on the data; and
circuitry that accesses the second memory based on the memory address to access one of the plurality of values.
15. The system of claim 14 , wherein the plurality of values are quotients.
16. The method of claim 1 , wherein:
determining the first scaling factor comprises computing, in real time, the first scaling factor in a first real-time computation based on the first data; and
computing the output data comprises computing, in real time, parameters for the second data in a second real-time computation based on the first scaling factor.
17. A method for scaling values in a real-time computation system, the method comprising:
receiving a first data comprising a first plurality of estimates of noise variance values;
determining, by a scaling factor computation circuit, a first scaling factor based on a smallest one of the first plurality of estimates of noise variance values;
receiving a second data after receiving the first data, wherein the second data comprises a second plurality of estimates of noise variance value; and
performing in parallel:
computing, by a computation circuit, output data based on the first scaling factor and the second data using a ratio of (i) the smallest one of the first plurality of estimates of noise variance values to (ii) one of the second plurality of estimates of noise variance values, and
determining a second scaling factor based on the second data, wherein computing the output data comprises computing detection parameters for the second data based on the smallest one of the first plurality of estimates of noise variance values associated with the first data.