Presentation is loading. Please wait.

Presentation is loading. Please wait.

DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi*, J. Zambreno.

Similar presentations


Presentation on theme: "DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi*, J. Zambreno."— Presentation transcript:

1 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi*, J. Zambreno +, G. Memik, and A. Choudhary Electrical Engineering and Computer Science Department *Indian Institute of Technology, Kharagpur, + Iowa State University Design, Automation & Test in Europe (DATE) 2008

2 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 2 Contributions  Novel architecture for Principal Component Analysis (PCA) used in Network Intrusion Detection  Parallel implementation of PCA on an FPGA platform  Achieving a throughput of 24.72 Gbps with up to 99.9% accuracy and as low as 1.95% false alarm rate Overview Principal Component Analysis PCA Framework Results

3 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 3  Network Attacks / Intrusions  Malicious Code (virus, worms, Trojan horses, malware)  Approximately $70 B economic losses a year (and increasing)  Example attacks: Denial of Service (DoS) Port scanning SYN Flood by exploiting TCP 3-way handshake Overview: Network Intrusions Overview Principal Component Analysis PCA Framework Results

4 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 4  Network Attacks / Intrusions  Malicious Code (virus, worms, Trojan horses, malware)  Approximately $70 B economic losses a year (and increasing)  Example attacks: Denial of Service (DoS) Port scanning Smurf attack Overview: Network Intrusions Overview Principal Component Analysis PCA Framework Results

5 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/20155 Overview: NIDS  Network Intrusion Detection System  Suspicious Activity Monitoring both inbound and outbound  Identify attacks and alert the System Administrator/User  NIDS techniques:  Signature Detection less false alarms, new attacks undetected (e.g. SNORT)  Anomaly Detection more false alarms, detection of new attacks  Popular Anomaly Detection Methods:  Data Mining, Genetic algorithms, Statistical analysis  Performance Bottlenecks: S/W methods cannot match Gigabit Ethernet rates Overview Principal Component Analysis PCA Framework Results

6 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 6 NIDS: Challenges  Problems with Signature IDS  Can only detect known intrusion types  Problems with Anomaly IDS  Need to have “sufficient” training data that covers as much variation of the normal behavior as possible  Too much dependence over training data set  False positives  Slow due to high dimensional network data Overview Principal Component Analysis PCA Framework Results

7 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 7 NIDS: Challenges  Problems with Signature IDS  Can only detect known intrusion types  Problems with Anomaly IDS  Need to have “sufficient” training data that covers as much variation of the normal behavior as possible  Too much dependence over training data set  False positives  Slow due to high dimensional network data Offline Phase Online Phase Labeled Training Data Learning Preprocessing Live Network Stream Deviation Attack Detection Overview Principal Component Analysis PCA Framework Results Normal Behavior

8 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary Overall NIDS Framework 6/25/2015 8 Overview Principal Component Analysis PCA Framework Results

9 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 9 Outline  Overview of Network Intrusions Detection  Principal Component Analysis (PCA)  Distance Calculation  Methodology  PCA Framework  PCA architecture  FPGA implementation  Results  FPGA performance  Accuracy  H/W Speedup  Conclusions Overview Principal Component Analysis PCA Framework Results

10 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 10 Principal Component Analysis (PCA)  Express variation in multivariate data in uncorrelated variables  Input data dimension consists of p correlated variables x 1,x 2,…x p  PCA looks for a transformation of the x i into p new variables y i that are uncorrelated y 1  f 1 ( x 1,x 2,…x p ) y 2  f 2 ( x 1,x 2,…x p ) ……… y p  f p ( x 1,x 2,…x p ) Eigenvalues ( λ i ) Eigenvectors ( e i ) Overview Principal Component Analysis PCA Framework Results

11 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary PCA: An Example Original Variable A Original Variable B PC 1 PC 2  PC 1 is the direction of maximum variance (major)  PC 2 uncorrelated (orthogonal) to PC 1 (minor) Overview Principal Component Analysis PCA Framework Results 6/25/2015 11

12 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 12 PCA: Properties  Principal components are arranged in descending order indicating the amount of variation it encompasses  Keep the q most important ones and discard the others  Some deviation may not follow the same correlation model  May want to keep the r least important components  Calculate distance in the new set of axes  Outlier if distance far from normal Overview Principal Component Analysis PCA Framework Results

13 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 13 y 1 y 2.. y q.. y p-r+1.. y p Major principal components Minor principal components PCA Overview Principal Component Analysis PCA Framework Results

14 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 14 PCA: Steps involved  Distance Calculation: Map the input data into suitable axes; calculate the distance using principal components  Euclidian Distance (equal weight for each dimension)  Mahalanobis Distance (uses correlation matrix)  Outlier Detection: Distance exceeding a pre- defined threshold  t M and t m with Major and Minor components, respectively Overview Principal Component Analysis PCA Framework Results

15 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/201515 Offline Phase Online Phase PCA Framework: Phases Overview Principal Component Analysis PCA Framework Results

16 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 16 PCA Framework: Phases Offline Phase:  Step 1: Mean vector of the sample data (training data) created  Step 2: Correlation Matrix R created  Step 3: Eigen-analysis performed on R; extraction of pairs  Step 4: Pairs are sorted according to λ i to determine the major and minor principal components  Step 5: Thresholds are calculated Overview Principal Component Analysis PCA Framework Results

17 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 17 PCA Framework: Phases Online Phase:  Step 1: Online data (test data) are mapped to the eigenspace of q & r principal components  Step 2: MajC and MinC scores are computed in parallel  Step 3: Threshold comparison is performed Overview Principal Component Analysis PCA Framework Results

18 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 18 PCA: Advantages  Reduces the dimensionality of network data with little or no information loss  Analyses different network TCP parameters (features)  Complete breakdown into offline and online phase  Eigen-analysis and sorting performed in offline phase  Major and minor components chosen define two thresholds  Better coverage  Faster compared to software implementation  Hardware level parallelism to compute PC score Overview Principal Component Analysis PCA Framework Results

19 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 19 Principal Component Score Pipeline for FPGA Overview Principal Component Analysis PCA Framework Results

20 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 20 FPGA Implementation  Design synthesis using VHDL  Place and route using Xilinx ISE 8.1  Target Platform Xilinx XC2VP30  # of pipeline stages = F(p, q, r)  Training and testing dataset :  KDD Cup 1999  100,000-125,000 training data samples  Up to 12 of 32 principal components used  Speedup limited by I/O bandwidth of FPGA board Overview Principal Component Analysis PCA Framework Results

21 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 21 Outline  Overview of Network Intrusions Detection  Principal Component Analysis (PCA)  Distance Calculation  Methodology  PCA Framework  PCA architecture  FPGA implementation  Results  FPGA performance  Accuracy  H/W Speedup  Conclusions Overview Principal Component Analysis PCA Framework Results

22 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 22 Results: FPGA Performance  # fields (p) = 32  Maximum throughput = 24.72 Gbps Overview Principal Component Analysis PCA Framework Results

23 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary Results: PCA Accuracy Detection and False Alarm Rates vs. q Overview Principal Component Analysis PCA Framework Results 6/25/2015 23

24 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 24 PCA hardware Speedup  Compared with sequential software code  System: AMD Opteron 2.4GHZ with 2GB memory Overview Principal Component Analysis PCA Framework Results

25 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 25 Conclusions  High Ethernet throughput demands efficient anomaly detection in networks  High dimensional network data becomes a bottleneck for anomaly detection  Hardware implementation of NIDS based on Principal Component Analysis can be effective  Parallelism is exploited using reconfigurable hardware (FPGAs)  Attack detection up to 99% with false alarm rates as low as 1.95%  Significant speedup compared to software Overview Principal Component Analysis PCA Framework Results

26 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary Acknowledgements This work was supported by NSF grants NSF-ITR CCR-0325207, CNS-0406341, CNS-0551639, IIS-0536994, CCR-0325207, by Air Force Office of Scientific Research (AFOSR) award FA9550-06-1- 0152 and DoE CAREER Award DE-FG02-05ER25691 6/25/2015 26

27 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 27 Thank You ! Questions? Contact Info: Electrical Engineering and Computer Science Department Northwestern University 2145 Sheridan Road Evanston, IL 60208 Phone: (847) 467-4610 Fax: (847) 467-4144 Email: ada829@eecs.northwestern.edu Web: http://www.ece.northwestern.edu/~ada829http://www.ece.northwestern.edu/~ada829

28 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 28  [1] M. E. Attig and J. Lockwood. A framework for rule processing in reconfigurable network systems. In IEEE Symposium on Field-Programmable Custom Computing Machines,(FCCM), Napa, CA, Apr. 2005.  [2] Z. K. Baker and V. K. Prasanna. Time and Area Efficient Pattern Matching on FPGAs. In The Twelfth Annual ACM International Symposium on Field-Programmable Gate Arrays (FPGA ’04), 2004  [3] Z. K. Baker and V. K. Prasanna. Efficient Hardware Data Mining with the Apriori Algorithm on FPGAs. In Proceedings of the Thirteenth Annual IEEE Sym. on Field Programmable Custom Computing Machines 2005  [4] S. Dharmapurikar, P. Krishnamurthy, T. Sproull, and J. W. Lockwood. Deep packet inspection using parallel bloom filters. In Symp. on High Performance Interconnects, August 2003.  [5] M. Fleury, B. Self, and A. C. Downton. A fine-grained parallel pipelined karhunen-loeve transform. In 17th International Parallel and Distributed Processing Symposium, Nice, France, April 2003.  [6] J. D. Jobson. Applied Multivariate Data Analysis, Volume II: Categorical and Multivariate Methods. Springer-Verlag, NY, 1992.  [7] I. T. Jolliffe. Principal Component Analysis. Springer-Verlag, NY, 2002. References

29 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 29  [8] Jungsuk Song and Hiroki Takakura and yasuo Okabe. A proposal of new benchmark data to evaluate mining algorithms for intrusion detection. In 23rd Asia Pacific Advanced Networking Meeting, 2007.  [9] KDD Cup 1999 Data. http://kdd.ics.uci.edu/databases/kddcup99/kddcup99.html, August 1999.  [10] Nicholas Athanasiades and Randal Abler and John Levine and Henry Owen and George Riley. Intrusion detection testing and benchmarking methodologies. In IEEE International Information Assurance Workshop, 2003.  [11] R. Sidhu and V. Prasanna. Fast regular expression matching FPGAs. In IEEE Symposium on Field-Programmable Custom Computing Machines, April 2001.  [12] D. V. Schuehler, J. Moscola, and J.W. Lockwood. Architecture for a hardware-based, tcp/ip content-processing system. In IEEE Micro, January 2004.  [13] M.-L. Shyu, S.-C. Chen, K. Sarinnapakorn, and L. Chang. A novel anomaly detection scheme based on principal component classifier. In IEEE Foundations and New Directions of Data Mining Workshop, pages 172–179, November 2003.  [14] H. Song and J. W. Lockwood. Efficient packet classification for network intrusion detection using fpga. In Intl. Symp. On Field-Programmable Gate Arrays, February 2005. References

30 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 30 Backups

31 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 31 NIDS FPGA Architecture Feature Extraction Principal Component Analysis Overview Principal Component Analysis PCA Framework Results

32 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 32 Feature Extraction Module HF(2) HF(1) HF(H) Estimate Block Hash Control Block 32 16 Src IP Dst IP Dst port Src port Sketch table H Sketch table 2 Sketch table 1 Estimate d value :::::: :::: :::: Flags 6 Feature Sketch Architecture

33 DATE 2008 A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System 6/25/2015 33

34 DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi, J. Zambreno, G. Memik and A. Choudhary 6/25/2015 34 Results: PCA Accuracy Detection and False alarm Rates vs. q Overview Principal Component Analysis PCA Framework Results


Download ppt "DATE 2008 An Efficient FPGA Implementation of Principal Component Analysis based Network Intrusion Detection System A. Das, S. Misra, S. Joshi*, J. Zambreno."

Similar presentations


Ads by Google