secure metering

35
Secure Metering Peng Wang

Upload: asha

Post on 12-Jan-2016

70 views

Category:

Documents


0 download

DESCRIPTION

Secure Metering. Peng Wang. Outline. Introduction An Auditable Metering Scheme for Web Advertisement Applications, ISC 01, Liqun Chen and Wenbo Mao Secure and efficient Metering, Eurocrypt’98, by Naor and Pinkas - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Secure Metering

Secure Metering

Peng Wang

Page 2: Secure Metering

Outline Introduction

1. An Auditable Metering Scheme for Web Advertisement Applications, ISC 01, Liqun Chen and Wenbo Mao

2. Secure and efficient Metering, Eurocrypt’98, by Naor and Pinkas

3. Third party certification of HTTP service access statistics, International Workshop on Security Protocols 99, F. Bergadano and P. De Mauro

4. SAWM: a tool for secure and authenticated web metering, SEKE’02, by Blundo and Cimato

Discussion

Page 3: Secure Metering

Introduction Metering schemes count (approximately) the number of

clients who visited a website Measure the popularity of website E.g. in order to decide on advertisement fees.

Secure against servers inflate the number of visits Secure against clients disrupt the metering process

Some schemes

Page 4: Secure Metering

Players

Web server

Clients

Request

Web page

setupsetup

Audit agency

proof

Request, page, evidence, etc.

1. Setup2. Evidence3. Proof

Page 5: Secure Metering

97

98

99

00

01

02

03

04

Timing function approachPolynomial approachTamper resistant hardware approachHash chain approach

Page 6: Secure Metering

97

98

99

00

01

02

03

04

Timing function approach

Auditable Metering with Lightweight Security, M. K. Franklin and D. Malkhi, Financial Crypto 97.

Use timing function to monitor the duration of browsing. Lightweight security that makes a large number of artificial visits very

costly.

An Auditable Metering Scheme for Web Advertisement Applications, Liqun Chen and Wenbo Mao, ISC 01

More details later.

Page 7: Secure Metering

97

98

99

00

01

02

03

04

Polynomial approach

Secure accounting and auditing on the Web, Moni Naor and Benny Pinkas, 7th international conference on World Wide Web

Secure and Efficient Metering, Moni Naor and Benny Pinkas, Eurocrypt '98

More details later.

Page 8: Secure Metering

97

98

99

00

01

02

03

04

Polynomial approach

Provably Secure Metering Scheme, Wakaha Ogata and Kaoru Kurosawa, ASIACRYPT '00 An attack on Noar and Pinkas paper, and the fix

Carlo Blundo and Annalisa De Bonis and Barbara Masucci and Douglas R. Stinson Metering Schemes with Pricing Dynamic Multi-threshold Metering Schemes A note on optimal metering schemes Bounds and constructions for metering schemes Metering Schemes for General Access Structures Efficient Metering Schemes With Pricing An information theoretic approach to metering schemes

Page 9: Secure Metering

97

98

99

00

01

02

03

04

Tamper resistant hardware approach

Third party certification of HTTP service access statistics, F. Bergadano and P. De Mauro, International Workshop on Security Protocols, 98

More details later.

Page 10: Secure Metering

97

98

99

00

01

02

03

04

Hash chain approach

SAWM: a tool for secure and authenticated web metering, Carlo Blundo and Stelvio Cimato, SEKE '02 More details later.

A Software Infrastructure for Authenticated Web Metering, Carlo Blundo and Stelvio Cimato, IEEE Computer

Page 11: Secure Metering

97

98

99

00

01

02

03

04

Lightweight security, timing function approachPolynomial approachTemper resistant hardware approachHash chain approach

Page 12: Secure Metering

An Auditable Metering Scheme Use timing function to monitor the duration of browsing a

web page that contains a ad bar. Lightweight security that makes a large number of

artificial visits very costly.

Setup: The audit agency generates e, n = qp, and sends (e,n) to

the website. Regular operation:

A web server sends the required web page with the ad bar, a timing function code and parameters (n, x, e) to clients upon requests.

x must be fresh The client’s browser displays the web page and runs the

timing function.

Page 13: Secure Metering

An Auditable Metering Scheme

The auditing algo takes 3 mod exponentiations.

When the client leaves the page, the code sends (t, a, x, e, n) to the server.

Server send (t, a, x, e, n) to the audit agency who runs:

Page 14: Secure Metering

+s & -s+ No client registration & modification+ Audit agency can be offline+ Can measure duration of visiting

- Light weight security- High verification overhead

- must verify each individual evidence

Some Server overhead

Page 15: Secure Metering

Secure and efficient Metering Based on Shamir’s polynomial secret sharing scheme Measure whether a server receives k visits in a certain

time frame Four schemes:

The basic scheme, A robust version, One that allows clients anonymity, One that allows unlimited use.

Page 16: Secure Metering

The basic scheme Initialization : (before any time frame)

A: generates a bivariate polynomial P(x, y) over Zp Degree k-1in x, and degree d-1 in y k: the expected number of visits d: the number of time frames The polynomial can be used < d time frames

A C: P(C, y)

Page 17: Secure Metering

The basic scheme cont

Regular operations: (when C visits S in the time frame t) C S: P(C, S || t) Assume that no two pairs (S || t) and (S’ || t’) are mapped to

the same element of Zp.

Proof generation: (after time frame t) If S has k different visitors in time frame t, it has k value:

{P(Ci, S || t)i=1…k}. S performs a Lagrange interpolation and computes the

proof: P(0, S || t)

S sends the proof to A who then verifies the proof.

Page 18: Secure Metering

+s & -s+ strong+ low verification overhead+ Audit agency can be offline

- Clients must register

Some server overhead

Page 19: Secure Metering

Third party certification of HTTP service access statistics

Setup: the website obtains a tamper-proof box from the audit agency and connects the box to a web server.

Regular operation: The web server inputs every client request to the box. The box generates random bits

The probability of 1 must be low to reduce overhead. If 0, the web server serves the client request normally. If 1,

1. the box Mac the request. 2. the web server redirects the request to the audit agency3. the audit agency verifies the request. 4. It redirects the request back to the web server who serves the

request. the web server log the operations

Verification: the web server submits the log to the audit agency who verifies the log file.

Page 20: Secure Metering

+s & -s+ No client registration & modification

- Require special hardware- Audit agency must be online- Server overhead & delay

Secure?

Page 21: Secure Metering

SAWM The authors did many work on polynomial based metering

schemes This paper describes a hash chain based solution Initialization:

For each client C, A generates a random value w and computes Hk(w)

A C: k, w A S: C, Hk(w)

C builds the hash chain: H(w)… Hk(w) S stores C, Hk(w) and sets a counter Lc to 0.

Page 22: Secure Metering

SAWM cont

Regular operations: C has the hash chain: H(w), H2(w) , … Hk-1(w), Hk(w) C S: Hk-j(w) for j-th visit S verifies and stores it, then increases the counter Lc

Proof generation and verification: SA: [C, Hk-Lc(w), Lc] for each client A has w and can verify.

Page 23: Secure Metering

+s & -s+ Low server overhead+ Audit agency can be offline

- Clients must register- High verification overhead

- must verify each individual evidence

Some Server overhead

Page 24: Secure Metering

No client registration & modification

Strong No online Agency

Can measure duration

Low server overhead

Low verification overhead

Timing function

Polynomial

Tamper proof hardware

Hash chain

Comparison

Page 25: Secure Metering

Discussion How to use metering schemes measure the

number of client visited an out-sourced service?

Page 26: Secure Metering

Secure Outsourcing

Yongdae Kim

Page 27: Secure Metering

Outsourcing Outsourcing is popular and will be much more popular

Examples ISP, SSP, DBSP Intrusion Detection Web service, Grid computing P2P…

Any outsourcing is involved with Service Level Agreement (SLA) Current SLA is vague, so that the service provider can fight

in case of dispute Otherwise, most outsourcing cases are based on trust

Page 28: Secure Metering

Outsourcing Primitives Counting

WS.com promised to provide web access to at least 10,000 users in 10 minute interval

Bandwidth Streaming.com promised to provide streaming service with

at least 100Kbps for any client of cnn.com Connection Computation Down-time Storage Database Intrusion detection

If your domain is shut down by the worm, secure.com will pay all cost associated with the worm to insecure.com.

Hybrid

Page 29: Secure Metering

Trustworthy Outsourcing Previous work focuses on mostly web counting

Research Theme Framework to develop a trustworthiness between a client

(service providee) and a server (service provider) in outsourcing

Simplification of dispute resolution

Page 30: Secure Metering

Trustworthy Outsourcing: Topics Extending web counting to more diverse primitives (as

outlined in previous slides) Fairness

A client can prove to a third party (or public) that the service was not provided, when it was not.

A server can prove to a third party (or public) that the service was provided, when it was.

How can we relax the assumption on participants? Auditing Agent? Client authentication required?

Fair P2P (file sharing) system

Page 31: Secure Metering

Supplemental

Page 32: Secure Metering

Robustness If clients send incorrect evidence, then the server cannot

compute the proof. The server must verify the evidence.

Given v=au+b, if S has (a, b), S can verify (v, u)

Initialization: A: Generates random polynomial P(x, y), A(x, y), and B(y).

Computes V(x, y) = A(x, y)*P(x, y) + B(y) A C: V(C, y) and P(C, y) A S: A(x, S || ti ) and B(S || ti ), i = 1…# of time frames

Page 33: Secure Metering

Robustness cont

Regular operations: C S: V(C, S || t) and P (C, S || t) S verifies if V = AP +B at the point (C, S || t)

Proof generation: S uses the k value {P(Ci, S || t)i=1…k} only to perform a

Lagrange interpolation and computes the proof: P(0, S || t)

S sends the proof to A who then verifies the proof.

Page 34: Secure Metering

Anonymity Sounds counterintuitive, but: Initialization: (based on basic scheme)

A: Generates random polynomial P(x, y), Qc(y) A C: P(Qc(y), y) and P(C, y)

Regular operations: C S: Qc(S || t) and P (Qc(S || t), S || t)

Proof generation: S performs a Lagrange interpolation and computes the proof:

P(0, S || t) S sends the proof to A who then verifies the proof.

Page 35: Secure Metering

A scheme for unlimited use g is a generator of a subgroup of Zp*, with order q Initialization:

A: Generates random polynomial P(x) of degree k-1 over Zq A C: P(C) and A’s signature of gP(C)

At the beginning of time frame t: A S: a challenge gr

Regular operations: S C: gr C S: A’s signature of gP(C), grp(c) , and a proof of the discrete log of

gP(C) to the base g and the discrete log of grP(C) to the base gr are the same.

Proof generation: SA: grp(0)