Malware clustering based on function call graph similarity

1. A method comprising:

maintaining, by a computer system, malware information corresponding to a plurality of malware samples;

receiving, by the computer system, information corresponding to a particular malware sample;

based at least in part on the information corresponding to the particular malware sample, generating, by the computer system, a function call graph corresponding to the particular malware sample;

generating, by the computer system based on the function call graph corresponding to the particular malware sample, a particular string representing the function call graph corresponding to the particular malware sample, wherein the particular string is indicative of a structure of the function call graph for the particular malware sample and identifies relationships between a plurality of function calls associated with the particular malware sample;

generating, by the computer system based at least in part on the particular string, a particular set of statistical properties associated with the particular malware sample;

executing, by the computer system, a query against the malware information to retrieve malware samples similar to the particular malware sample, wherein the query specifies one or more of the particular set of statistical properties;

based on the query, identifying, by the computer system, a first subset of malware samples, of the plurality of malware samples, that match the particular malware sample; and

assigning, by the computer system, the particular malware sample and the first subset of malware samples into one of a plurality of clusters of related malware samples.

2. The method of claim 1 , wherein the information corresponding to the particular malware sample includes code for the particular malware sample, and wherein generating the particular string comprises:

identifying the plurality of function calls, wherein the plurality of function calls are included in the code for the particular malware sample; and

generating the function call graph for the particular malware sample based on the relationships between the plurality of function calls.

3. The method of claim 2 , wherein the code for the particular malware sample is provided in an assembly language, and wherein the code specifies names for the plurality of function calls in each of the plurality of malware samples.

4. The method of claim 1 , wherein the particular set of statistical properties includes a distribution of function calls included in the function call graph for the particular malware sample, and wherein executing the query is based on the particular set of statistical properties.

5. The method of claim 1 , wherein the particular string is an adjacency list specifying a call structure of the plurality of function calls called by the particular malware sample.

6. The method of claim 1 , further comprising storing, by the computer system, the particular string and the particular set of statistical properties as part of the malware information.

7. The method of claim 6 , further comprising:

receiving, by the computer system, subsequent information corresponding to a subsequent malware sample;

based at least in part on the subsequent information corresponding to the subsequent malware sample, generating, by the computer system, a function call graph corresponding to the subsequent malware sample;

generating, by the computer system based on the function call graph corresponding to the subsequent malware sample, a string representing the function call graph corresponding to the subsequent malware sample;

generating, by the computer system based at least in part on the string representing the function call graph corresponding to the subsequent malware sample, a set of statistical properties associated with the subsequent malware sample;

executing, by the computer system, a subsequent query against the malware information; and

based on the subsequent query, identifying, by the computer system, a second subset of malware samples, of the plurality of malware samples, that match the subsequent malware sample, wherein the second subset of malware samples includes the particular malware sample.

8. The method of claim 1 , further comprising analyzing, by the computer system, statistical properties associated with the first subset of malware samples to identify at least one anomalous sample included in the first subset of malware samples, wherein, during the assigning, the at least one anomalous sample is not included in the one of the plurality of clusters of related malware samples.

9. A non-transitory, computer-readable medium having instructions stored thereon that, when executed by a processor of a computer system, cause the computer system to perform operations comprising:

maintaining malware information corresponding to a plurality of malware samples;

receiving, from a client system, information corresponding to a particular malware sample;

based at least in part on the information corresponding to the particular malware sample, generating a function call graph corresponding to the particular malware sample;

generating, based on the function call graph corresponding to the particular malware sample, a particular string representing the function call graph corresponding to the particular malware sample, wherein the particular string is indicative of a structure of the function call graph for the particular malware sample and identifies relationships between a plurality of function calls associated with the particular malware sample;

generating, based at least in part on the particular string, a particular set of statistical properties associated with the particular malware sample;

executing a query against the malware information to retrieve malware samples similar to the particular malware sample, wherein the query specifies one or more of the particular set of statistical properties;

based on the query, identifying a subset of malware samples, of the plurality of malware samples, that match the particular malware sample; and

assigning the particular malware sample and the subset of malware samples into one of a plurality of clusters of related malware samples.

10. The non-transitory, computer-readable medium of claim 9 , wherein the operations further comprise sending, to the client system, information specifying that the particular malware sample has been assigned to the one of the plurality of clusters of related malware samples, wherein the information further specifies a malware classification of the one of the plurality of clusters of related malware samples.

11. The non-transitory, computer-readable medium of claim 9 , wherein the operations further comprise generating, for a first malware sample of the plurality of malware samples, a first string specifying a function call graph for the first malware sample, and wherein generating the first string specifying the function call graph for the first malware sample comprises:

accessing a directed graph associated with the first malware sample;

identifying a plurality of function calls included in the directed graph;

generating the function call graph for the first malware sample based on relationships between the plurality of function calls included in the directed graph; and

generating the first string based on the function call graph for the first malware sample, wherein the first string is indicative of a structure of the function call graph for the first malware sample, and wherein the first string is included in the malware information.

12. The non-transitory, computer-readable medium of claim 9 , wherein identifying the subset of malware samples is performed using a subgraph-matching algorithm.

13. A system comprising:

at least one processor; and

a memory having instructions stored thereon that are executable by the at least one processor to cause the system to perform operations comprising

maintaining malware information corresponding to a plurality of malware samples,

receiving information corresponding to a particular malware sample,

based at least in part on the information corresponding to the particular malware sample, generating a function call graph corresponding to the particular malware sample,

generating, based on the function call graph corresponding to the particular malware sample, a particular string representing the function call graph corresponding to the particular malware sample, wherein the particular string is indicative of a structure of the function call graph for the particular malware sample and identifies relationships between a plurality of function calls associated with the particular malware sample,

generating, based at least in part on the particular string, a particular set of statistical properties associated with the particular malware sample,

executing a query against the malware information to retrieve malware samples similar to the particular malware sample, wherein the query specifies one or more of the particular set of statistical properties,

based on the query, identifying a first subset of malware samples, of the plurality of malware samples, that match the particular malware sample, and

assigning the particular malware sample and the first subset of malware samples into one of a plurality of clusters of related malware samples.

14. The system of claim 13 , wherein the information corresponding to the particular malware sample includes code for the particular malware sample, and wherein generating the particular string comprises:

identifying the plurality of function calls, wherein the plurality of function calls are included in the code for the particular malware sample; and

generating the function call graph for the particular malware sample based on the relationships between the plurality of function calls.

15. The system of claim 14 , wherein the code for the particular malware sample is provided in an assembly language, and wherein the code specifies names for the plurality of function calls in each of the plurality of malware samples.

16. The system of claim 13 , wherein the particular set of statistical properties includes a distribution of function calls included in the function call graph for the particular malware sample, and wherein executing the query is based on the particular set of statistical properties.

17. The system of claim 13 , wherein the particular string is an adjacency list specifying a call structure of the plurality of function calls called by the particular malware sample.

18. The system of claim 13 , wherein the operations further comprise storing the particular string and the particular set of statistical properties as part of the malware information.

19. The system of claim 18 , wherein the operations further comprise:

receiving subsequent information corresponding to a subsequent malware sample;

based at least in part on the subsequent information corresponding to the subsequent malware sample, generating a function call graph corresponding to the subsequent malware sample;

generating, based on the function call graph corresponding to the subsequent malware sample, a string representing the function call graph corresponding to the subsequent malware sample;

generating, based at least in part on the string representing the function call graph corresponding to the subsequent malware sample, a set of statistical properties associated with the subsequent malware sample;

executing a subsequent query against the malware information; and

based on the subsequent query, identifying a second subset of malware samples, of the plurality of malware samples, that match the subsequent malware sample, wherein the second subset of malware samples includes the particular malware sample.

20. The system of claim 13 , wherein the operations further comprise analyzing statistical properties associated with the first subset of malware samples to identify at least one anomalous sample included in the first subset of malware samples, wherein, during the assigning, the at least one anomalous sample is not included in the one of the plurality of clusters of related malware samples.