Network-based and Attack-resilient Length Signature Generation for Zero-day Polymorphic Worms

Network-based and Attack-resilient Length Signature Generation for Zero-day Polymorphic Worms

Zhichun Li1, Lanjia Wang2, Yan Chen1 and Judy Fu3

1 Lab for Internet and Security Technology (LIST), Northwestern Univ.

2 Tsinghua University, China

3 Motorola Labs, USA

2

The Spread of Sapphire/Slammer Worms

3

Limitations of Exploit Based Signature

1010101

10111101

11111100

00010111

Our network

Traffic Filtering

Internet

Signature: 10.*01

XX

Polymorphic worm might not have exact exploit based signature

Polymorphism!

4

Vulnerability Signature

Work for polymorphic wormsWork for all the worms which target thesame vulnerability

Vulnerability signature traffic filtering

Internet

XX Our network

UnknownVulnerability

XX

Better!

5

Benefits of Network Based Detection

• At the early stage of the worm, only limited worm samples.

• Host based sensors can only cover limited IP space, which might have scalability issues.

Gateway routersInternet

Our network

Host baseddetection

Early Detection!

6

Design Space and Related Work

• Most host approaches depend on lots of host information, such as source/binary code of the vulnerable program, vulnerability condition, execution traces, etc.

[Polygraph-SSP05][Hamsa-SSP06][PADS-INFOCOM05]

[CFG-RAID05]

[Nemean-Security05]

[DOCODA-CCS05]

[TaintCheck-NDSS05]

LESG (this paper)

[Vulsig-SSP06]

[Vigilante-SOSP05]

[COVERS-CCS05]

[ShieldGen-SSP07]

Vulnerability Based

Exploit Based

Network Based Host Based

7

Outline

• Motivation and Related Work

• Design of LESG

• Problem Statement

• Three Stage Algorithm

• Attack Resilience Analysis

• Evaluation

• Conclusions

8

Basic Ideas

• At least 75% vulnerabilities are due to buffer overflow

• Intrinsic to buffer overflow vulnerability and hard to evade

• However, there could be thousands of fields to select the optimal field set is hard

Vulnerable buffer

Protocol message

Overflow!

9

FrameworkProtocolClassifier

UDP1434

LESGWormFlow

Classifier

TCP137

. . .TCP80

TCP53

TCP25

NormalTraffic Pool

SuspiciousTraffic Pool

Signatures

NetworkTap

KnownWormFilter

Normal traffic reservoir

Real time

Policy driven

ICDCS06, INFOCOM06, TON

10

LESG Signature Generator

11

Outline

• Motivation and Related Work

• Design of LESG

• Problem Statement

• Three Stage Algorithm

• Attack Resilience Analysis

• Evaluation

• Conclusions

12

Field Hierarchies

DNS PDU

13

Length-based Signature Definition

Name Type Class TTL RDlength RDATA

Length Signature (Name,100)

100

Length Signature

Signature Set{(Name,100), (Class,50), (RDATA,300)}

“OR” relationship

Ground truth signature(RDATA,315)

Buffer length!

Vulnerable

RDATA

14

Problem Formulation

LESG

Worms which are not covered in the suspicious pool are at most

Minimize the false positives in the normal pool

Suspicious pool

Normal pool

Signature

With noise NP-Hard!

15

Outline

• Motivation and Related Work

• Design of LESG

• Problem Statement

• Three Stage Algorithm

• Attack Resilience Analysis

• Evaluation

• Conclusions

16

Stages I and II

Stage I: Field Filtering Stage II: Length Optimization

COV≥1%FP≤0.1%

Trade off between specificity and sensitivityScore function Score(COV,FP)

17

Stage III• Find the optimal set of fields as the

signature with high coverage and low false positive

• Separate the fields to two sets, FP=0 and FP>0– Opportunistic step (FP=0)– Attack Resilience step (FP>0)

• The similar greedy algorithm for each step

18

Stage III (cont.)

suspicious normal

Name Type Class TTL Comments RDATA

(Name,100) [40%,0.03%]

(Class,50) [35%,0.09%]

(RDATA,300) [50%,0.05%]

(Comments,2000) [10%,0.1%]

Stage I COV0≥1%FP0≤0.1%

50% 0.05% Residual coverage≥5%

19

Stage III (cont.)

suspicious normal

Name Type Class TTL Comments RDATA

(Class,50) [25%,0.02%]

(Name,100) [3%,0.08%]

{(RDATA,300)}

(Comments,2000) [1%,0.05%]

Stage I COV0≥1%FP0≤0.1%

50% 0.05% Residual coverage≥5%

20

Stage III (cont.)

suspicious normal

Name Type Class TTL Comments RDATA

(Class,50) [25%,0.02%]

(Name,100) [3%,0.08%]

{(RDATA,300),(Class,50)}

(Comments,2000) [1%,0.05%]

Stage I COV0≥1%FP0≤0.1%

(50+25)%(0.05+0.02)% Residual coverageγ≥5%

21

Attack Resilience Bounds

• Depend on whether deliberated noise injection (DNI) exists, we get different bounds

• With 50% noise in the suspicious pool, we can get the worse case bound FN<2% and FP<1%

• In practice, the DNI attack can only achieve FP<0.2%

• Resilient to most proposed attacks (proposed in other papers)

22

Outline

• Motivation and Related Work

• Design of LESG

• Problem Statement

• Three Stage Algorithm

• Attack Resilience Analysis

• Evaluation

• Conclusions

23

Methodology• Protocol parsing with Bro and BINPAC (I

MC2006)

• Worm workload– Eight polymorphic worms created based on r

eal world vulnerabilities including CodeRed II and Lion worms.

– DNS, SNMP, FTP, SMTP

• Normal traffic data– 27GB from a university gateway and 123GB

email log

24

Results• Single/Multiple worms with noise

– Noise ratio: 0~80%– False negative: 0~1% (mostly 0)– False positive: 0~0.01% (mostly 0)

• Pool size requirement– 10 or 20 flows are enough even with 20% noises

• Speed results– With 500 samples in suspicious pool and 320K s

amples in normal pool, For DNS, parsing 58 secs, LESG 18 secs

25

Conclusions

• A novel network-based automated worm signature generation approach– Work for zero day polymorphic worms with

unknown vulnerabilities – First work which is both Vulnerability based

and Network based using length signature for buffer overflow vulnerabilities

– Provable attack resilience– Fast and accurate through experiments