Granted Patent B2

US 12,489,544 · App. 18/537,619 · Granted Dec 2, 2025

Inventors: Aladin Djuhera (Dachau, DE); Swanand Ravindra Kadhe (San Jose, CA); Farhan Ahmed (San Jose, CA); Fernando Luiz Koch (Palm Beach Gardens, FL); Heiko H. Ludwig (San Francisco, CA); Alecio Pedro Delazari Binotto (Munich, DE)

Assignee: International Business Machines Corporation

H04K3/22H04L63/1416

Abstract

Claims (80)

1 . A computer-implemented method (CIM), the CIM comprising:

performing a detecting and monitoring adversarial jamming (MDJ) process for an edge environment that includes a first edge device and a central hub;

in response to a determination, based on results of performing the MDJ process, that a jamming attack event is likely to occur, causing a pre-emptive edge user device protection process (PPP) to be deployed for preventing the jamming attack event;

in response to a determination, based on the results of performing the MDJ process, that a jamming attack event has occurred, causing a post-process jammed data (PJD) process to be deployed at a recipient of a wireless transmission subject to the jamming attack event;

generating, based on results of deploying the PPP and/or the PJD process, a first set of new parameters for a model of a device that sent the wireless transmission subject to the jamming attack event to the recipient; and

causing the first set of new parameters to be distributed to the model of the device that sent the wireless transmission subject to the jamming attack event to the recipient.

2 . The CIM of claim 1 , wherein the edge environment includes a plurality of edge devices including the first edge device, wherein performing the MDJ process for the edge environment includes:

evaluating wireless transmissions within the edge environment;

identifying, based on results of the evaluation, anomalies that have occurred during the wireless transmissions;

determining, based on the identified anomalies, the edge devices that are potentially under threat of jamming attack events and a potential range of the jamming attack events; and

constructing, for each of the edge devices, a probability mapping, wherein each of the probability mappings detail, for an associated one of the edge devices, a likelihood of the jamming attack events affecting the associated edge device.

3 . The CIM of claim 2 , wherein the probability mappings includes insights about the jamming attack events, wherein the insights are selected from the group consisting of: behavior of a malevolent device that initiates at least one of the jamming attack events, signal strength of a wireless transmission subject to the jamming attack event, and a list of the edge devices that are threatened by the jamming attack event, and further comprising:

generating, based on results of deploying the PPP and/or the PJD process, sets of new parameters for models of the edge devices; and

causing the sets of new parameters to be distributed to the edge devices.

4 . The CIM of claim 1 , wherein deployment of the PPP includes:

determining a suitable number of redundant samples to be added to data that is to be wirelessly transmitted between the first edge device and the central hub;

determining an intentional noise level to add to the data;

applying randomized smoothing during wireless transmission of the data between the first edge device and the central hub; and

applying data clustering in response to a determination that the data has been wirelessly transmitted between the first edge device and the central hub.

5 . The CIM of claim 4 , wherein applying data clustering includes identifying portions of the data that qualify as outliers from a remainder of the data, and instructing a recipient of the data wirelessly transmitted between the first edge device and the central hub to ignore the identified portions of the data.

6 . The CIM of claim 1 , wherein deployment of the PJD process includes:

determining an optimal clustering of data of the wireless transmission subject to the jamming attack event;

discarding portions of the data that are identified during the determination of the optimal clustering to qualify as outliers; and

determining a quantification of a cleansing impact that performing the PJD process has on the data of the wireless transmission subject to the jamming attack event.

7 . The CIM of claim 6 , wherein determining the optimal clustering of the data of the wireless transmission subject to the jamming attack event includes:

evaluating potential clustering guidance from previously performed data-centric analysis;

determining clustering parameters to use for determining the optimal clustering of the data;

selecting a clustering algorithm to use for determining the optimal clustering of the data; and

causing the selected clustering algorithm to run on the data of the wireless transmission subject to the jamming attack event, wherein portions of the data identified by the selected clustering algorithm to lie outside of determined clusters are identified to qualify as outliers.

8 . The CIM of claim 1 , wherein generating the first set of new parameters for the model of the device that sent the wireless transmission subject to the jamming attack event to the recipient includes averaging current model parameters with weighted coefficients.

9 . The CIM of claim 8 , wherein the weighted coefficients are based on a determined quantification of a cleansing impact that performing the PJD process has on data of the wireless transmission subject to the jamming attack event or are based on additions and smoothing incorporated into the data of the wireless transmission subject to the jamming attack event.

10 . A computer program product (CPP), the CPP comprising:

a set of one or more computer-readable storage media;

program instructions, collectively stored in the set of one or more storage media, for causing a processor set to perform the following computer operations:

perform a detecting and monitoring adversarial jamming (MDJ) process for an edge environment that includes a first edge device and a central hub;

in response to a determination, based on results of performing the MDJ process, that a jamming attack event is likely to occur, cause a pre-emptive edge user device protection process (PPP) to be deployed for preventing the jamming attack event;

in response to a determination, based on the results of performing the MDJ process, that a jamming attack event has occurred, cause a post-process jammed data (PJD) process to be deployed at a recipient of a wireless transmission subject to the jamming attack event;

generate, based on results of deploying the PPP and/or the PJD process, a first set of new parameters for a model of a device that sent the wireless transmission subject to the jamming attack event to the recipient; and

cause the first set of new parameters to be distributed to the model of the device that sent the wireless transmission subject to the jamming attack event to the recipient.

11 . The CPP of claim 10 , wherein the edge environment includes a plurality of edge devices including the first edge device, wherein performing the MDJ process for the edge environment includes:

evaluating wireless transmissions within the edge environment;

identifying, based on results of the evaluation, anomalies that have occurred during the wireless transmissions;

determining, based on the identified anomalies, the edge devices that are potentially under threat of jamming attack events and a potential range of the jamming attack events; and

constructing, for each of the edge devices, a probability mapping, wherein each of the probability mappings detail, for an associated one of the edge devices, a likelihood of the jamming attack events affecting the associated edge device.

12 . The CPP of claim 11 , wherein the probability mappings includes insights about the jamming attack events, wherein the insights are selected from the group consisting of: behavior of a malevolent device that initiates at least one of the jamming attack events, signal strength of a wireless transmission subject to the jamming attack event, and a list of the edge devices that are threatened by the jamming attack event, and further comprising:

program instructions, collectively stored in the set of one or more storage media, for causing the processor set to perform the following computer operations:

generate, based on results of deploying the PPP and/or the PJD process, sets of new parameters for models of the edge devices; and

cause the sets of new parameters to be distributed to the edge devices.

13 . The CPP of claim 10 , wherein deployment of the PPP includes:

determining a suitable number of redundant samples to be added to data that is to be wirelessly transmitted between the first edge device and the central hub;

determining an intentional noise level to add to the data;

applying randomized smoothing during wireless transmission of the data between the first edge device and the central hub; and

applying data clustering in response to a determination that the data has been wirelessly transmitted between the first edge device and the central hub.

14 . The CPP of claim 13 , wherein applying data clustering includes identifying portions of the data that qualify as outliers from a remainder of the data, and instructing a recipient of the data wirelessly transmitted between the first edge device and the central hub to ignore the identified portions of the data.

15 . The CPP of claim 10 , wherein deployment of the PJD process includes:

determining an optimal clustering of data of the wireless transmission subject to the jamming attack event;

discarding portions of the data that are identified during the determination of the optimal clustering to qualify as outliers; and

determining a quantification of a cleansing impact that performing the PJD process has on the data of the wireless transmission subject to the jamming attack event.

16 . The CPP of claim 15 , wherein determining the optimal clustering of the data of the wireless transmission subject to the jamming attack event includes:

evaluating potential clustering guidance from previously performed data-centric analysis;

determining clustering parameters to use for determining the optimal clustering of the data;

selecting a clustering algorithm to use for determining the optimal clustering of the data; and

causing the selected clustering algorithm to run on the data of the wireless transmission subject to the jamming attack event,

wherein portions of the data identified by the selected clustering algorithm to lie outside of determined clusters are identified to qualify as outliers.

17 . The CPP of claim 10 , wherein generating the first set of new parameters for the model of the device that sent the wireless transmission subject to the jamming attack event to the recipient includes averaging current model parameters with weighted coefficients.

18 . The CPP of claim 17 , wherein the weighted coefficients are based on a determined quantification of a cleansing impact that performing the PJD process has on data of the wireless transmission subject to the jamming attack event or are based on additions and smoothing incorporated into the data of the wireless transmission subject to the jamming attack event.

19 . A computer system (CS), the CS comprising:

a processor set;

a set of one or more computer-readable storage media;

program instructions, collectively stored in the set of one or more storage media, for causing the processor set to perform the following computer operations:

perform a detecting and monitoring adversarial jamming (MDJ) process for an edge environment that includes a first edge device and a central hub;

in response to a determination, based on results of performing the MDJ process, that a jamming attack event is likely to occur, cause a pre-emptive edge user device protection process (PPP) to be deployed for preventing the jamming attack event;

in response to a determination, based on the results of performing the MDJ process, that a jamming attack event has occurred, cause a post-process jammed data (PJD) process to be deployed at a recipient of a wireless transmission subject to the jamming attack event;

generate, based on results of deploying the PPP and/or the PJD process, a first set of new parameters for a model of a device that sent the wireless transmission subject to the jamming attack event to the recipient; and

cause the first set of new parameters to be distributed to the model of the device that sent the wireless transmission subject to the jamming attack event to the recipient.

20 . The CS of claim 19 , wherein the edge environment includes a plurality of edge devices including the first edge device, wherein performing the MDJ process for the edge environment includes:

evaluating wireless transmissions within the edge environment;

identifying, based on results of the evaluation, anomalies that have occurred during the wireless transmissions;

determining, based on the identified anomalies, the edge devices that are potentially under threat of jamming attack events and a potential range of the jamming attack events; and

constructing, for each of the edge devices, a probability mapping, wherein each of the probability mappings detail, for an associated one of the edge devices, a likelihood of the jamming attack events affecting the associated edge device.

Assignments (1)

Continuity (1)

References Cited (21)