惡意代碼檢測

惡意代碼定義

惡意代碼也稱爲惡意軟件，是對各種敵對和入侵軟件的概括性術語。包括各種形式的計算機病毒、蠕蟲、特洛伊木馬、勒索軟件、間諜軟件、廣告軟件以及其他的惡意軟件。

惡意代碼的種類

計算機病毒：指寄居在計算機系統中，在一定條件下被執行會破壞系統、程序的功能和數據，影響系統其他程序和自我複製。

蠕蟲：也算是一種病毒，它具有自我複製能力並通過計算和網絡的負載，消耗有限資源。

特洛伊木馬：也可以簡稱爲木馬，最初來源於古希臘傳說。計算機木馬是一種潛伏在計算機中爲了達到某種特殊目的的程序，比如竊取用戶私密信息和控制用戶系統等。它與病毒最大的不同點在於，病毒能進行自我複製，而木馬不具有複製功能，不會感染其他程序。

Rootkit：最初是指一組能幫助使用者獲取系統權限的工具包，這裏的是一種惡意程序，用於獲取目標主機權限之後隱藏攻擊者訪問痕跡，使得攻擊者不被發現從而能夠長期擁有管理員權限。它具有很好的隱蔽性和潛伏性，難以檢測。

惡意代碼特徵（區分程序惡意特徵的特徵信息）

• 系統調用特徵
• 規範化代碼特徵
• N-gram特徵
• 控制流（CFG特徵）
• 指令序列特徵
• 文件格式等特徵

惡意代碼特徵提取

Byte n-gram Features:從文件的二進制代碼中提取Byte n-gram特徵，其中選擇訓練集中每個類的L個最常出現的n克來表示類的配置文件。

Opcode n-gram Features:首先拆卸所有數據集的可執行文件和操作碼提取。一個操作碼的彙編語言指令描述要執行的操作。它是短形式的操作碼。一條指令包含一個操作碼和操作數,選擇應該採取的操作。一些操作的操作數操作碼可能操作,根據CPU體系結構,寄存器,值存儲在內存和堆棧等等。一個操作碼的作用在算術、邏輯運算和數據處理操作。操作碼能夠統計得出之間的可變性惡意和正版軟件。

Portable Executables：這些特徵是從EXE文件的某些部分提取出來的。利用可執行文件的結構信息，通過靜態分析提取可執行文件的特徵。這些有意義的特性表明文件被操縱或感染以執行惡意活動。

String Features：這些特徵是基於純文本編碼在可執行文件，如windows, getversion, getstartupinfo, getmodulefilename, messagebox，庫等。這些字符串是用PE和非PE可執行文件編碼的連續可打印字符。

Function Based Features：在程序文件的運行時行爲上提取基於函數的特徵。基於函數的特性函數駐留在要執行的文件中，並利用它們生成表示文件的各種屬性。

Hybrid Analysis Features：靜態分析和動態分析的結合。

惡意代碼檢測

• 基於靜態特徵的惡意代碼檢測技術

分類特徵	參考文獻
The byte code	Kolter J Z, Maloof M A. Learning to detect malicious executables in the wild. [C]. Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining. ACM, 2004: 470-478. Santos I, Penya Y K, Devesa J, et al. N-grams-based File Signatures for Malware Detection. Proceedings of the 2009 International Conference on Enterprise Information Systems (ICEIS), 2009, 9: 317-320
n-grams
File format	Shafiq M Z, Tabish S M, Mirza F, et al. Pe-miner: Mining structural information to detect malicious executables in realtime. Recent advances in intrusion detection, Springer Berlin Heidelberg, 2009: 121-141. Bai J, Wang J, Zou G. A Malware Detection Scheme Based on Mining Format Information. The Scientific World Journal, 2014.
Gray image	Nataraj L, Karthikeyan S, Jacob G, et al. Malware images: visualization and automatic classification[C] . Proceedings of the 8th international symposium on visualization for cyber security. ACM, 2011: 4. HAN Xiao-guang, QU Wu, YAO Xuan-xia, et al. Research on malicious code variants detection based on texture fingerprint. Journal on Communications, 2014, 35(8):125-135.
Function call graph	Kong D, Yan G. Discriminant malware distance learning on structural information for automated malware classification[C]. Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining. ACM, 2013: 1357-1365.

• 基於動態特徵的惡意代碼檢測技術

分類特徵	參考文獻
Variable length	Nair V P, Jain H, Golecha Y K, et al. MEDUSA: MEtamorphic malware dynamic analysis usingsignature from API[C]. Proceedings of the 3rd International Conference on Security of Information and Networks. ACM, 2010: 263-269. Chen F, Fu Y. Dynamic detection of unknown malicious executables base on api interception[C]. Database Technology and Applications, 2009 First International Workshop on. IEEE, 2009: 329-332. Firdausi I, Lim C, Erwin A, et al. Analysis of machine learning techniques used in behavior-based malware detection[C]. Advances in Computing, Control and Telecommunication Technologies (ACT), 2010 Second International Conference on. IEEE, 2010: 201-203.
API	Nair V P, Jain H, Golecha Y K, et al. MEDUSA: MEtamorphic malware dynamic analysis usingsignature from API[C]. Proceedings of the 3rd International Conference on Security of Information and Networks. ACM, 2010: 263-269.
subsequences
Operation code	Shabtai A, Moskovitch R, Feher C, et al. Detecting unknown malicious code by applying classification techniques on opcode patterns. Security Informatics, 2012, 1(1): 1-22. [17] Pai S, Di Troia F, Visaggio C A, et al. Clustering for malware classification. Journal of Computer Virology and Hacking Techniques, 2016: 1-13.
n-grams
Graph	Bonfante G, Kaczmarek M, Marion J Y. Architecture of a morphological malware detector. Journal in Computer Virology, 2009, 5(3): 263-270. Cesare S, Xiang Y, Zhou W. Control flow-based malware variant detection. IEEE Transactions on Dependable and Secure Computing, 2014, 11(4): 307–317.

• 基於融合特徵的惡意代碼檢測技術（各種集成特徵類型的檢測方法）

分類特徵（動態特徵/靜態特徵）	參考文獻
Dynamic API operation code	SantosI, DevesaJ, Brezo F, et al. Opem: A static-dynamic approach for machine learning based malware detection[C]. International Joint Conference CISIS’12-ICEUTE´ 12-SOCO´ 12 Special Sessions. Springer Berlin Heidelberg, 2013: 271-280
Program behavior Static DLL、API	Lu Y B, Din S C, Zheng C F, et al. Using multi-feature and classifier ensembles to improve malware detection. Journal of CCIT, 2010, 39(2): 57-72.
API call sequence PE format	Guo S, Yuan Q, Lin F, et al. A malware detection algorithm based on multi-view fusion. Neural Information Processing, Models and Applications, Springer Berlin Heidelberg, 2010: 259-266. Krawczyk B, Woźniak M. Evolutionary Cost-Sensitive Ensemble for Malware Detection. International Joint Conference SOCO’14-CISIS’14-ICEUTE’14, Springer International Publishing, 2014: 433-442.
Dynamic API Static API	Ozdemir M, Sogukpinar I. An Android Malware Detection Architecture based on Ensemble Learning. Transactions on Machine Learning and Artificial Intelligence, 2014, 2(3): 90-106.
operation code byte code	Bai, Jinrong, and Junfeng Wang. Improving malware detection using multiview ensemble learning. Security and Communication Networks 9.17 (2016): 4227-4241.

參考文獻：

[1] Bo Yun Zhang.Survey on Malicious Code Intelligent Detection Techniques

[2]Smita Ranveer,Swapnaja Hiray.Comparative Analysis of Feature Extraction Methods of Malware Detection