Amplification of formal method and fuzz testing to enable scalable assurance for communication system
View Patent ↗Methods for more secure mobile network communications are disclosed. Specifically, details involving natural language processing (NLP) based auto formal modeling of protocols and specifications with large language models (NLP) are provided. Methods for formal and fuzzing amplification for fuzz testing to detect vulnerabilities are also disclosed. Furthermore, solutions are provided to identified vulnerabilities in existing 5G infrastructures. Also disclosed is a digital twin fuzzing framework.
1 . A method for validating communication systems, comprising steps of:
obtaining input data comprising protocol descriptions;
transforming said protocol descriptions into dependency graphs and formal models using a large language model, said large language model being integrated with a transformer model;
resolving ambiguities in said input data;
determining design intent in said input data;
obtaining iterative feedback from a HyFuzz platform, said HyFuzz platform being adapted to refine capture of intentions and resolve ambiguities in said input data by providing said input data to said HyFuzz platform, wherein said HyFuzz platform is configured to supply said design intent to said large language model; and
producing output.
2 . The method of claim 1 , further comprising a step of establishing dependency relationships through cross-attention mechanisms and/or self-attention mechanisms.
3 . The method of claim 1 , further comprising a step of quantifying dependency relationships.
4 . The method of claim 3 , wherein said dependency relationships are quantified without human involvement.
5 . The method of claim 1 , further comprising a step of creating a list of exploits from said input data.
6 . The method of claim 5 , further comprising a step of conducting fuzz testing on said list of exploits.
7 . The method of claim 1 , wherein said large language model is Cross Attention Learning.
8 . The method of claim 1 , wherein said large language model performs concurrent processing of said protocol descriptions.
9 . The method of claim 1 , wherein said dependency graphs are scalable cross-session dependency graphs, supporting hierarchy of formal analysis, thereby facilitating revelation of in-depth relationships embedded within said protocol descriptions.
10 . The method of claim 1 , wherein said HyFuzz platform is trained on labeled identifiers and formal properties.
11 . The method of claim 1 , wherein said output bypasses token count barriers.
12 . A method for validating communication systems, comprising steps of:
obtaining input data comprising protocol descriptions;
transforming said protocol descriptions into dependency graphs and formal models using a large language model, said large language model being integrated with a transformer model:
resolving ambiguities in said input data;
determining design intent in said input data;
obtaining iterative feedback from an experimental platform by providing said input data to said experimental platform, wherein said experimental platform is configured to supply said design intent to said large language model; and
producing output, wherein said output bypasses token count barriers.
13 . The method of claim 12 , further comprising a step of establishing dependency relationships through cross-attention mechanisms and/or self-attention mechanisms.
14 . The method of claim 13 , further comprising a step of quantifying said dependency relationships.
15 . The method of claim 14 , wherein said dependency relationships are quantified without human involvement.
16 . The method of claim 12 , further comprising a step of creating a list of exploits from said input data.
17 . The method of claim 16 , further comprising a step of conducting fuzz testing on said list of exploits.
18 . The method of claim 12 , wherein said large language model performs concurrent processing of said protocol descriptions.
19 . The method of claim 12 , wherein said dependency graphs are scalable cross-session dependency graphs, supporting hierarchy of formal analysis, thereby facilitating revelation of in-depth relationships embedded within said protocol descriptions.
20 . The method of claim 12 , wherein the experimental platform is trained on labeled identifiers and formal properties.
21 . The method of claim 12 , wherein said large language model is Cross Attention Learning.
22 . A method for validating communication systems, comprising steps of:
obtaining input data comprising protocol descriptions;
transforming said protocol descriptions into dependency graphs and formal models using a large language model, said large language model being integrated with a transformer model;
resolving ambiguities in said input data;
determining design intent in said input data;
obtaining iterative feedback from an experimental platform by providing said input data to said experimental platform, wherein said experimental platform is configured to supply said design intent to said large language model;
producing output; and
determining if identifiers from said protocol descriptions inherit formal properties from previous identifiers of said protocol descriptions, thereby identifying direct and indirect dependency relationships.
23 . The method of claim 22 , wherein said direct and indirect dependency relationships are established through cross-attention mechanisms and/or self-attention mechanisms.
24 . The method of claim 22 , further comprising a step of quantifying said direct and indirect dependency relationships.
25 . The method of claim 24 , wherein said direct and indirect dependency relationships are quantified without human involvement.
26 . The method of claim 22 , further comprising a step of creating a list of exploits from said input data.
27 . The method of claim 26 , further comprising a step of conducting fuzz testing on said list of exploits.
28 . The method of claim 22 , wherein said large language model performs concurrent processing of said protocol descriptions.
29 . The method of claim 22 , wherein said output bypasses token count barriers.
30 . The method of claim 22 , wherein said dependency graphs are scalable cross-session dependency graphs, supporting hierarchy of formal analysis, thereby facilitating revelation of in-depth relationships embedded within said protocol descriptions.
31 . The method of claim 22 , wherein the experimental platform is trained on labeled identifiers and formal properties.
32 . The method of claim 22 , wherein said large language model is Cross Attention Learning.