IP Library Granted Patent US 12,632,775
Granted Patent B2
US 12,632,775 · App. 17/180,247 · Granted May 19, 2026

Machine learning supplemented storage device calibration

Inventors: Jonathan Lloyd (Newport Beach, CA); Anand Gupta (Bengaluru, IN); Stella Achtenberg (Netanya, IL); Ofir Pele (Hod HaSharon, IL); Chun Sei Tsai (Tustin, CA); Amit Chattopadhyay (San Jose, CA); Aimamorn Suvichakorn (Nonthaburi, TH); Krzysztof Gladysz (Katowice, PL); Kameron Jung (Yorba Linda, CA)
Assignee: Western Digital Technologies, Inc.
G06N20/00G06N5/04G11B19/048
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,632,775
App. No.
17/180,247
Granted
May 19, 2026
Kind
B2
Abstract

Methods are provided for deploying machine learning operations within existing storage devices for streamlining various calibration processes. Machine learning operations are specifically designed to generate inference data as a substitute for various measurements taken during calibration. These operations may be verified through additional sample measurements and rolled back when the results of the machine learning operations are outside of a range of approved values. Storage devices designed to utilize machine learning methods within calibration processes can include a non-volatile memory for storing data, executable instructions, and a processor to conduct a variety of steps. The steps can include executing an application stored in the non-volatile memory and receiving a request for measurement data from the application. The steps can further determine if the requested data is suitable for substitution by an inference and subsequently select at least one machine learning model for generating a suitable inference.

Claims (48)

1 . A storage device, comprising:

a Non-Volatile Memory (NVM) for storing data; and

one or more processors communicatively coupled to the NVM, the one or more processors being configured, individually or in combination, to direct the storage device to:

execute within the storage device an application configured to generate measurement data for a plurality of attributes associated with calibration of the storage device to account for physical variations unique to the storage device;

generate first measurement data by internally measuring at least one attribute of the plurality of attributes;

based on the first measurement data, dynamically generate within the storage device one or more thresholds corresponding to the at least one attribute;

based on the first measurement data, internally generate inference data for remaining attributes of the plurality of attributes via an embedded machine-learning model;

generate second measurement data by internally measuring a computationally-inexpensive attribute from the remaining attributes of the plurality of attributes;

based at least in part on the dynamically-generated one or more thresholds, internally determine whether inference data for the computationally-inexpensive attribute corresponds to the second measurement data; and

responsive to the inference data for the computationally-inexpensive attribute corresponding to the second measurement data, utilize the inference data for the remaining attributes in place of measurement data to avoid use of resources associated with directly obtaining the measurement data;

wherein each of the measurement data and the inference data are utilized to include generation of calibration data corresponding to a bit length corresponding to a change of magnetic flux over a given area of a track of a recording disk surface of a hard disk drive storage device to minimize the area required to store a bit on the recording disk surface.

2 . The storage device of claim 1 , wherein the application is further configured to compare the second measurement data against the inference data for the computationally-inexpensive attribute to determine whether the inference data for the computationally-inexpensive attribute is within a range of acceptable values.

3 . The storage device of claim 2 , wherein the application is further configured to not utilize the inference data responsive to the comparison between the second measurement data and the inference data for the computationally-inexpensive attribute being outside of the range of acceptable values.

4 . The storage device of claim 3 , wherein the application is further configured to measure and utilize the remaining attributes responsive to the comparison between the second measurement data and the inference data for the computationally-inexpensive attribute being outside of the range of acceptable values.

5 . The storage device of claim 4 , wherein the one or more processors are further configured, individually or in combination, to direct the storage device to store within a data log non-utilized data associated with the inference data and associated processing.

6 . The storage device of claim 2 , wherein the range of acceptable values is based on the dynamically-generated one or more thresholds.

7 . The storage device of claim 2 , wherein:

the measurement of attributes comprises varying levels of computational resources to complete; and

the computationally-inexpensive attribute measured is chosen based on the level of computational resources required.

8 . The storage device of claim 1 , wherein the calibration data is utilized to determine an areal density capability for the hard disk drive storage device for allocating a format corresponding to a target capacity for each recording disk surface of the hard disk drive storage device.

9 . The storage device of claim 1 , wherein at least one of the plurality of attributes is related to at least one of: heads, zones, tracks, or sectors.

10 . The storage device of claim 1 , wherein the application is further configured to internally evaluate an entire data set of the measurement data and the inference data to verify validity.

11 . A method for calibrating a storage device based on machine-learning based inferences within a storage device, comprising:

executing within the storage device an application configured to generate measurement data for each of a plurality of attributes associated with calibration of the storage device to account for variations unique to the particular storage device;

generating first measurement data by internally measuring one attribute of the plurality of attributes;

based on the first measurement data, dynamically generating within the storage device one or more thresholds corresponding to the one attribute;

based on the first measurement data, internally generating inference data for remaining attributes of the plurality of attributes via an embedded machine-learning model;

internally generating second measurement data by measuring a computationally-inexpensive attribute from the remaining attributes of the plurality of attributes;

based at least in part on the dynamically-generated one or more thresholds, internally determining whether inference data for the computationally-inexpensive attribute corresponds to the second measurement data; and

responsive to the inference data for the computationally-inexpensive attribute corresponding to the second measurement data, utilizing the inference data for the remaining attributes in place of the measurement data to avoid use of resources associated with directly obtaining the measurement data;

wherein each of the measurement data and the inference data is utilized to include generation of calibration data corresponding to a bit length corresponding to a change of magnetic flux over a given area of a track of a recording disk surface of a hard disk drive storage device, corresponding to a particular head configured to operate over the recording disk surface, to minimize the area required to store a bit on the recording disk surface.

12 . The method of claim 11 , further comprising comparing the second measurement data against the inference data for the computationally-inexpensive attribute to determine whether the inference data for the computationally-inexpensive attribute is within a range of acceptable values corresponding to the dynamically-generated one or more thresholds.

13 . The method of claim 12 , further comprising not utilizing the inference data responsive to the comparison between the second measurement data and the inference data for the computationally-inexpensive attribute being outside of the range of acceptable values.

14 . The method of claim 13 , further comprising measuring and utilizing the remaining attributes responsive to the comparison between the second measurement data and the inference data for the computationally-inexpensive attribute being outside of the range of acceptable values.

15 . A storage device, comprising:

means for storing data and executable instructions; and

one or more processors communicatively coupled to the means for storing, the one or more processors being configured, individually or in combination, to direct the storage device to:

receive a request from an application process for measurement of a plurality of attributes associated with calibration of hard disk drive storage device formatting to account for physical variations unique to the particular hard disk drive storage device;

internally generate first measurement data by measuring at least one attribute of the plurality of attributes;

based on the first measurement data, dynamically generate within the storage device one or more thresholds corresponding to the at least one attribute;

based on the first measurement data, internally generate inference data for remaining attributes of the plurality of attributes via an embedded machine-learning model;

internally generate second measurement data by measuring a computationally-inexpensive attribute from the remaining attributes of the plurality of attributes;

based at least in part on the dynamically-generated one or more thresholds, internally determine whether inference data for the computationally-inexpensive attribute corresponds to the second measurement data; and

responsive to the inference data for the computationally-inexpensive attribute corresponding to the second measurement data, utilize the inference data for the remaining attributes in place of measurement data to avoid use of resources associated with directly obtaining measurement data to generate calibration data corresponding to a flux change per inch over a given track of a recording disk surface of a hard disk drive storage device to minimize the area required to store a bit on the recording disk surface.

16 . The storage device of claim 15 , wherein the one or more processors are further configured, individually or in combination, to direct the storage device to compare the second measurement data against the inference data for the computationally-inexpensive attribute to determine whether the inference data for the computationally-inexpensive attribute is within a range of acceptable values corresponding to the dynamically-generated one or more thresholds.

17 . The storage device of claim 16 , wherein the one or more processors are further configured, individually or in combination, to direct the storage device to not utilize the inference data responsive to the comparison between the second measurement data and the inference data for the computationally-inexpensive attribute being outside of the range of acceptable values.

18 . The storage device of claim 17 , wherein the one or more processors are further configured, individually or in combination, to direct the storage device to measure and utilize the remaining attributes responsive to the comparison between the second measurement data and the inference data for the computationally-inexpensive attribute being outside of the range of acceptable values.

19 . The storage device of claim 15 , wherein the one or more processors are further configured, individually or in combination, to internally evaluate an entire data set of the inference data to verify validity.

Assignments (5)
PATENT COLLATERAL AGREEMENT - A&R LOAN AGREEMENT Recorded Aug 21, 2023
From: WESTERN DIGITAL TECHNOLOGIES, INC.
To: JPMORGAN CHASE BANK, N.A.
Reel/Frame 064715/0001 →
PATENT COLLATERAL AGREEMENT - DDTL LOAN AGREEMENT Recorded Aug 21, 2023
From: WESTERN DIGITAL TECHNOLOGIES, INC.
To: JPMORGAN CHASE BANK, N.A.
Reel/Frame 067045/0156 →
RELEASE OF SECURITY INTEREST AT REEL 056285 FRAME 0292 Recorded Feb 8, 2022
From: JPMORGAN CHASE BANK, N.A.
To: WESTERN DIGITAL TECHNOLOGIES, INC.
Reel/Frame 058982/0001 →
SECURITY INTEREST Recorded May 19, 2021
From: WESTERN DIGITAL TECHNOLOGIES, INC.
To: JPMORGAN CHASE BANK, N.A., AS AGENT
Reel/Frame 056285/0292 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Feb 19, 2021
From: LLOYD, JONATHAN; GUPTA, ANAND; ACHTENBERG, STELLA; PELE, OFIR; TSAI, CHUN SEI; CHATTOPADHYAY, AMIT; SUVICHAKORN, AIMAMORN; GLADYSZ, KRZYSZTOF; JUNG, KAMERON
To: WESTERN DIGITAL TECHNOLOGIES, INC.
Reel/Frame 055339/0936 →
Continuity (2)
Provisional Application 63094088 · Oct 20, 2020
Related Publication 20220121985A1 · Apr 21, 2022
References Cited (78)
US 6725098B2 · Edwards · 2004 [cited by examiner]
US 7707461B2 · Dougherty · 2010 [cited by examiner]
US 7930507B2 · Perry · 2011 [cited by examiner]
US 8489525B2 · Bisdikian · 2013 [cited by examiner]
US 9305595B2 · Hwang · 2016 [cited by examiner]
US 10108791B1 · Masterman · 2018 [cited by examiner]
US 10176829B1 · Li · 2019 [cited by examiner]
US 10310821B2 · Brown · 2019 [cited by examiner]
US 10339448B2 · Guo · 2019 [cited by examiner]
US 10430719B2 · David · 2019 [cited by examiner]
US 10438620B1 · Ellis · 2019 [cited by examiner]
US 10719239B2 · Rowstron · 2020 [cited by examiner]
US 10748582B1 · Feist · 2020 [cited by examiner]
US 10777224B2 · Ellis · 2020 [cited by examiner]
US 10810512B1 · Wubbels · 2020 [cited by examiner]
US 10817441B2 · Kumar · 2020 [cited by examiner]
US 10831396B2 · Fisher · 2020 [cited by examiner]
US 10861320B2 · Martin · 2020 [cited by examiner]
US 10901628B2 · Das · 2021 [cited by examiner]
US 10963394B2 · Kachare · 2021 [cited by examiner]
US 10990092B2 · Ghosh · 2021 [cited by examiner]
US 11243711B2 · Sato · 2022 [cited by examiner]
US 11257002B2 · Faulhaber, Jr. · 2022 [cited by examiner]
US 11295209B2 · Thiruvengadam · 2022 [cited by examiner]
US 11320165B2 · Venkatesh · 2022 [cited by examiner]
US 11435924B2 · Akutsu · 2022 [cited by examiner]
US 11449796B2 · Song · 2022 [cited by examiner]
US 11503615B2 · Hu · 2022 [cited by examiner]
US 11507860B1 · Hodgman · 2022 [cited by examiner]
US 11521017B2 · Abelha Ferreira · 2022 [cited by examiner]
US 11531831B2 · Jia · 2022 [cited by examiner]
US 11587156B2 · Morin · 2023 [cited by examiner]
US 11592984B2 · Tiotantra · 2023 [cited by examiner]
US 11734143B2 · Raz · 2023 [cited by examiner]
US 11790766B2 · Martin · 2023 [cited by examiner]
US 11861467B2 · Gadelrab · 2024 [cited by examiner]
US 11932274B2 · Woo · 2024 [cited by examiner]
US 12366970B2 · Lee · 2025 [cited by examiner]
US 20030078680A1 · Edwards · 2003 [cited by examiner]
US 20080180084A1 · Dougherty · 2008 [cited by examiner]
US 20100169603A1 · Perry · 2010 [cited by examiner]
US 20110289029A1 · Bisdikian · 2011 [cited by examiner]
US 20130132638A1 · Horn et al. · 2013 [cited by applicant]
US 20150221333A1 · Hwang · 2015 [cited by examiner]
US 20160125292A1 · Seo · 2016 [cited by examiner]
US 20170109646A1 · David · 2017 [cited by examiner]
US 20180053401A1 · Martin · 2018 [cited by examiner]
US 20180197574A1 · Guo · 2018 [cited by examiner]
US 20180349103A1 · Brown · 2018 [cited by examiner]
US 20190156247A1 · Faulhaber, Jr. · 2019 [cited by examiner]
US 20190179547A1 · Szubbocsev · 2019 [cited by applicant]
US 20190295579A1 · Ellis · 2019 [cited by examiner]
US 20190317901A1 · Kachare · 2019 [cited by examiner]
US 20190354283A1 · Rowstron · 2019 [cited by examiner]
US 20190385640A1 · Ellis · 2019 [cited by examiner]
US 20200192595A1 · Fisher · 2020 [cited by examiner]
US 20200293200A1 · Das · 2020 [cited by examiner]
US 20200301595A1 · Akutsu · 2020 [cited by examiner]
US 20200310993A1 · Kumar · 2020 [cited by examiner]
US 20200387152A1 · Ghosh · 2020 [cited by examiner]
US 20210097329A1 · Jia · 2021 [cited by examiner]
US 20210174238A1 · Song · 2021 [cited by examiner]
US 20210192333A1 · Thiruvengadam · 2021 [cited by examiner]
US 20210199328A1 · Venkatesh · 2021 [cited by examiner]
US 20210204300A1 · Hu · 2021 [cited by examiner]
US 20210233388A1 · Martin · 2021 [cited by examiner]
US 20210240381A1 · Sato · 2021 [cited by examiner]
US 20210279635A1 · Gadelrab · 2021 [cited by examiner]
US 20210319354A1 · Raz · 2021 [cited by examiner]
US 20210334597A1 · Abelha Ferreira · 2021 [cited by examiner]
US 20210357783A1 · Cogan et al. · 2021 [cited by applicant]
US 20210366031A1 · Morin · 2021 [cited by examiner]
US 20210380127A1 · Woo · 2021 [cited by examiner]
US 20220083230A1 · Tiotantra · 2022 [cited by examiner]
US 20220188028A1 · Mesnier · 2022 [cited by examiner]
US 20220405635A1 · Zhang · 2022 [cited by examiner]
US 20230147773A1 · Lee · 2023 [cited by examiner]
US 20240054384A1 · Catron · 2024 [cited by examiner]