Iterative Autocorrelation Calculation for Streamed Data

The patent segments the autocorrelation calculation into distinct components: mean calculation, variance calculation, and covariance calculation. Each component is computed separately and then combined to produce the final autocorrelation result. This segmentation allows for optimized computation of each part and enables parallel processing, thereby improving calculation efficiency while reducing overall computational load.

The patent performs preliminary calculations of the mean and variance before computing the covariance. By pre-computing these statistical parameters and storing them, the method avoids redundant calculations during the autocorrelation computation phase. This preliminary action significantly reduces the computational burden, especially when dealing with large buffered datasets.

The patent divides the large computation window into smaller manageable segments or blocks. Instead of loading the entire large window into memory at once, the method processes data in smaller chunks, computing statistical parameters for each segment and then aggregating the results. This approach maintains statistical accuracy over the full computation window while significantly reducing the buffer size required at any given time.

The patent implements a nested processing structure where smaller computation windows or blocks are nested within the larger computation window. Each nested block is processed independently, and their results are combined to produce the final autocorrelation for the entire large window. This nesting strategy enables accurate analysis of large datasets with limited memory resources.

The patent establishes continuous computation streams where statistical parameters (mean, variance, covariance) are updated incrementally as new data arrives, rather than performing complete recalculations. This continuous update mechanism maintains real-time responsiveness by always having current statistical values ready, while minimizing computational overhead through incremental updates instead of frequent full recalculations.

The patent pre-computes and maintains running statistics (mean, variance) that are updated continuously as data flows through the system. These pre-computed parameters are readily available when autocorrelation calculations are needed, eliminating the need for expensive full recalculations and enabling rapid response to new data while keeping power consumption low.

The patent segments the data access pattern by organizing computations around pre-calculated statistical parameters rather than accessing individual data elements. Instead of visiting all n data elements for each autocorrelation calculation, the method accesses the already-computed mean, variance, and covariance parameters, which encapsulate the information from all data elements. This ensures calculation completeness while dramatically reducing processing time.

The patent creates and uses copies of statistical parameters (mean, variance, covariance) that represent the entire data window. Rather than accessing the original data elements repeatedly, the system works with these computed parameter copies, which contain the essential statistical information. This copying approach maintains calculation reliability while minimizing data access time and computational complexity.