Vector Computing Unit for CDMA Descrambling and Despreading
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional ASIC-based implementations for multiple communication modes face challenges such as large chip area, long market orientation periods, high circuit customization, inflexibility, and high power consumption, especially in soft-baseband-based vector computing units for CDMA de-scrambling and de-spreading operations.
Innovation Solution
A method and vector computing unit that perform complex multiplication and data accumulation using data transformation and addition, storing results in a vector register file, and utilizing commands like mult 1nj and sxu to efficiently accumulate data, reducing the need for general multiplication and minimizing power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional ASIC manner is adopted for implementing multiple communication modes, then implementation reliability is improved, but chip area increases and market orientation period lengthens
Solution Approach 1:
The patent implements a universal soft-baseband processing platform that can handle multiple communication modes (GSM, WCDMA, LTE, etc.) through a single configurable architecture. The vector computing unit with configurable parameters (vector length, number of rows, accumulation depth) serves multiple functions across different communication standards, eliminating the need for separate ASICs for each mode and thereby reducing chip area while maintaining implementation reliability.
Solution Approach 2:
The patent employs dynamic configurability in the soft-baseband processing unit, where parameters such as vector length, number of rows, and accumulation depth can be adjusted based on the specific communication mode requirements. This dynamic adaptation allows the same hardware resource to be efficiently utilized across different scenarios, reducing the overall chip area compared to static ASIC implementations for each mode.
2Reliability
If conventional ASIC manner is adopted for implementing multiple communication modes, then implementation reliability is improved, but circuit customization degree increases and flexibility decreases
Solution Approach 1:
The soft-baseband processing platform provides universal support for multiple communication modes through configurable parameters. The vector computing unit can be programmed to implement different communication standards by adjusting parameters such as vector length and accumulation depth, thereby maintaining high flexibility and adaptability while ensuring reliable implementation across all modes without requiring high circuit customization degree.
Solution Approach 2:
The patent utilizes parameter changes to adapt the processing unit to different communication modes. By modifying parameters such as vector length, number of rows, and accumulation depth, the same hardware can reliably implement different communication standards, maintaining both reliability and flexibility without requiring extensive circuit customization.
3Adaptability or versatility
If soft-baseband-based vector computing unit is used for CDMA de-scrambling and de-spreading, then flexibility is improved, but processing efficiency decreases and power consumption increases
Solution Approach 1:
The patent segments the de-scrambling and de-spreading process into distinct stages: complex multiplication stage and accumulation stage. The vector computing unit processes data in configurable vector lengths with specific number of rows and accumulation depths. This segmentation allows optimized processing for each stage, improving overall efficiency while maintaining the flexibility of soft-baseband implementation.
Solution Approach 2:
The patent optimizes processing efficiency by configuring specific parameters for the vector computing unit: vector length, number of rows, and accumulation depth. These parameter changes enable the flexible soft-baseband implementation to achieve high processing efficiency by matching the computational requirements of CDMA de-scrambling and de-spreading operations.
4Adaptability or versatility
If soft-baseband-based vector computing unit is used for CDMA de-scrambling and de-spreading, then flexibility is improved, but power consumption increases
Solution Approach 1:
The patent segments the computational workload into distinct stages (complex multiplication and accumulation) with configurable vector lengths and rows. This segmentation allows the power-efficient vector computing unit to process operations in optimized batches, reducing overall power consumption compared to traditional soft-baseband approaches while maintaining flexibility.
Solution Approach 2:
The patent reduces power consumption by optimizing parameters such as vector length, number of rows, and accumulation depth. These parameter changes enable the flexible soft-baseband implementation to minimize computational overhead and power consumption by processing operations in efficiently configured batches.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
Provided are a method and vector computing unit for implementing de-scrambling and de-spreading, and a computer storage medium. The method includes that: an operation of complex multiplication on baseband data and corresponding de-scrambling and de-spreading codes is performed by adopting data transformation and addition, and data obtained by the complex multiplication is stored into a vector register file; a row of data obtained by the complex multiplication is read from the vector register file, every two adjacent pieces of data in the row of data are accumulated to obtain a half row of data, and the obtained data is stored back into the vector register file; a previous accumulation result in the vector register file or a previous accumulation result reserved in a result register of a vector computing unit is read, every two adjacent pieces of data in the previous accumulation result are accumulated to obtain a 1/4 row of data, and accumulation is continued to finally obtain one piece of accumulated data; and the accumulation processing is continued on other rows of data to implement accumulation of each row of data obtained by the complex multiplication in the vector register file.