1: % Paper Submission: July 1, 2003 (extended)
2: % Author Notification: Sep. 15, 2003
3: % Final Manuscript: Oct. 7, 2003
4: % $Id: main.tex,v 1.5 2003/06/16 13:47:04 s-yamane Exp s-yamane $
5: \documentclass[10pt,twocolumn]{article}
6: %\usepackage{latex8}
7: \usepackage{times}
8: \usepackage{hyperref} % URL encoding.
9: % ftp://ftp.tex.ac.uk/tex-archive/macros/latex2e/contrib/supported/hyperref/
10: \usepackage[dvips]{graphicx}
11: \begin{document}
12: \title{Rethinking OpenPGP PKI and OpenPGP Public Keyserver}
13: \author{
14: Shinji Yamane\\
15: Iwate Prefectural University\\
16: Faculty of Software and Information Science\\
17: Takizawa-aza-Sugo, Takizawa vil. Iwate 0200193 JAPAN\\
18: {\url{s-yamane@soft.iwate-pu.ac.jp}}
19: \and
20: Jiahong Wang\\ Iwate Prefectural University\\
21: pgp-folks@comm.soft.iwate-pu.ac.jp
22: \and
23: Hironobu Suzuki\\ Independent Software Consultant\\
24: pgp-folks@comm.soft.iwate-pu.ac.jp
25: \and
26: Norihisa Segawa\\ Iwate Prefectural University\\
27: pgp-folks@comm.soft.iwate-pu.ac.jp
28: \and
29: Yuko Murayama\\ Iwate Prefectural University\\
30: Murayama@iwate-pu.ac.jp
31: }
32: \maketitle
33: \thispagestyle{empty}
34:
35: \begin{abstract}
36: %TITLE: ``Rethinking OpenPGP PKI and OpenPGP Public Keyserver''\\
37: OpenPGP, an IETF Proposed Standard based on
38: PGP\textregistered\ application, has its own Public
39: Key Infrastructure (PKI) architecture which is different from the one
40: based on X.509, another standard from ITU. This paper describes
41: the OpenPGP PKI; the historical perspective as well as its current use.
42: We also compare three PKI technologies standardized by IETF:
43: OpenPGP, PKIX(X.509), and SPKI/SDSI.
44: %Finally we propose a new design and implementation for the next
45: %generation OpenPGP public keyserver.
46:
47: Since the OpenPGP PKI works without a registration authority nor
48: certification authority, it fits well with the Internet communication
49: with voluntary community.
50: For example, the digital signature for email including the security
51: patch program of free software is usually signed by not an authorized
52: organization but the cross-PGP-signed individuals who belong to
53: different organizations or nations.
54:
55: The current OpenPGP PKI issues include the capability of a PGP keyserver and
56: its performance. PGP keyservers have been developed and operated
57: by volunteers since the 1990s. The keyservers distribute, merge, and
58: expire the OpenPGP public keys.
59: Major keyserver managers from several countries have built the globally
60: distributed network of PGP keyservers. However, the current PGP Public
61: Keyserver (pksd) has some limitations. It does not support fully the
62: OpenPGP format so that it is neither expandable nor flexible, without
63: any cluster technology.
64:
65: Finally we introduce the project on the next generation OpenPGP
66: public keyserver called the OpenPKSD, lead by Hironobu Suzuki, one of the
67: authors, and funded by Japanese Information-technology Promotion Agency(IPA).
68: %The OpenPKSD is developed by Hironobu Suzuki
69: %and has been distributed its developers' version under GPL.
70: \end{abstract}
71: \setcounter{tocdepth}{4}
72: %\tableofcontents
73:
74: %-------------------------------------------------------------------------
75: \section{Introduction}
76:
77: Authentication is an essential factor of information security in network
78: society.
79: The difficulty of building a Public-Key Infrastructure (PKI) is a major
80: impediment to strong authentication.
81: Without PKI, we cannot trust neither digital signature nor certification
82: based on the public key cryptosystem via wide-area network.
83:
84: In following section \ref{sec:PKI} and \ref{sec:PKIs}, we overview the
85: PKI architecture comparing several models. In
86: section \ref{web_of_trust}, we examine a PKI without authorities which
87: is presented by the OpenPGP technology and compare it with the other
88: models.
89: % for Pros and Cons.
90: Then we point out the role of the PGP
91: keyserver in section \ref{sec:keyserver}
92: and introduce the next generation OpenPGP public keyserver
93: project in section \ref{OpenPKSD}.
94: Finally we give some conclusions.
95:
96: \section{PKI architectures}\label{sec:PKI}
97: PKI has three core functions as follows to manage
98: the users' certification and trust relations
99: ~\cite[s.v. ``public-key infrastructure'']{FYI36}:
100: \begin{enumerate}
101: \item to register users and issue their public-key certificates
102: \item to revoke certificates when required
103: \item to archive data needed to validate certificates at a much later
104: time
105: \end{enumerate}
106: %
107: To operate these three functions with many users on a large-scale network,
108: many PKI have a hierarchical structure for CAs
109: and are built using a centralized architecture.
110: However there are alternatives.
111:
112: PKI is categorized by the architecture types as follows:
113: 1) hierarchical PKI, 2) mesh PKI, and 3) trust-file PKI~\cite[s.v. ``trusted certificate'']{FYI36}.
114: The difference is the way how they rely on CA (Certification Authority).
115: A hierarchical PKI has the most significant CA in terms of trust
116: at the root of the hierarchy tree.
117: A mesh PKI has CAs issue cross-certificates to each other.
118: A trust-file PKI has a local file of public-key
119: certificates that the user trusts as starting points
120: for certification chain.
121:
122: For example, popular browsers are distributed with an initial file of
123: trusted certificates, the starting points for certification paths.
124: The initial file is different between among the each PKI
125: architecture.
126: In a hierarchical PKI, the initial file is the root certificate in a
127: hierarchical PKI. It is usually ``baked into'' the browsers with no
128: decisions by the users to trust them. In a mesh PKI, it is the
129: certificate of the CA that issued the user's own certificate.
130: And in a trust-file PKI, any
131: certificates including self-signed certificates
132: accepted by the user can be the first public key in a certification path.
133:
134: \section{PKI standards}\label{sec:PKIs}
135: To build PKIs, different standards are developed.
136: They are based on their own framework and architecture and they are never
137: the same.
138: This section compares different PKI architectures:
139: 1) X.509 standard from ITU,
140: 2) OpenPGP, an IETF Proposed standard based on PGP\textregistered\
141: application,
142: and
143: 3) SPKI, another standards based on the theoretical research.
144:
145: %\subsection{X.509-based PKI}
146: X.509 is the earliest framework
147: to provide and support authentication
148: including formats for X.509 public-key
149: certificates, X.509 attribute certificates, and X.509 CRLs.
150: X.509 is the hierarchical PKI that a CA, central digital certificates
151: issuer, is responsible for managing the certificates.
152: % CA$B$N5!G=$K$OEPO?5!G=$b4^$^$l$k$,!$$=$N5!G=$OEPO?5!4X(B (RA)$B$K0QBw$5$l$k>l9g(B
153: % $B$b$"$j!$I,$:$7$b(BCA$B<+$i$,EPO?$r9T$J$&I,MW$O$J$$!%(B
154:
155: Historically, X.509 was standardized by ITU-T (Inter\-national
156: Tele\-communication Union Tele\-communication sector, formerly CCITT) and
157: turned to be ISO standard.
158: X.509 follows the X.500 directory service and provides an example of
159: reliable authentication and certification.
160: In practice, developers relax the strict X.500 service scheme.
161: For example,
162: X.509v3 (Version 3) certificate has ``extensions'' field for
163: flexible operation.
164: % (see table \ref{table:X.509v3})$B!%(B
165: %
166: %\onecolumn
167: %\begin{table}[h]
168: % \begin{center}
169: % \begin{tabular}{|l|l|}
170: % \hline
171: % 1. version & Identifies v1, v2, or v3. \\
172: % 2. serialNumber & Certificate serial number; \\
173: % & an integer assigned by the issuer. \\
174: % 3. signature & OID (object identifier) of algorithm \\
175: % & that was used to sign the certificate. \\
176: % 4. issuer & DN (distinguished name) of the \\
177: % & issuer (the CA who signed). \\
178: % 5. validity & Validity period; a pair of UTCTime \\
179: % & values: ``not before'' and ``not after''. \\
180: % 6. subject & DN of entity who owns the public key. \\
181: % 7. subjectPublicKeyInfo & Public key value and algorithm OID. \\
182: % 8. issuerUniqueIdentifier & Defined for v2, v3; optional. \\
183: % 9. subjectUniqueIdentifier & Defined for v2, v2; optional. \\
184: % 10. extensions & Defined only for v3; optional. \\
185: % \hline
186: % \end{tabular}
187: % \end{center}
188: % \caption{X.509 public-key certificate}\label{table:X.509v3}
189: %\end{table}
190: %\twocolumn
191:
192: IETF had discussed about the design based on
193: X.509 framework from each applications to general PKI.
194: % S/MIME (Secure/Multipurpose Internet Mail Extensions) and
195: % PKCS (Public-Key Cryptography Standards)~\cite{RFC2986} proposed by
196: % RSA Data Security and standardized by
197: % IETF S/MIME Working Group
198: % ({\url{http://www.ietf.org/html.charters/smime-charter.html}}).
199: %
200: % S/MIME$B$NG'>Z5!9=$O(BPKIX Working Group$B$N(B
201: % $B:n6H$K4p$E$$$F9T$J$o$l$k$3$H$K$J$C$F$*$j!$N>%0%k!<%W$OL)@\$J4X78$K$"$k!%(B
202: % S/MIME$B$K$D$$$F$O(BRFC 2311, 2312$B$K$F5,3J2=$5$l!$$=$l$rF'$^$($?(B
203: % $B%$%s%?!<%M%C%H$G$N(BX.509$B$K$D$$$F$O(BRFC 2459$B$K$F5,3J2=$5$l$F$$$k!%(B
204: Internet standards for X.509 PKI framework is developed
205: at IETF Public-Key Infrastructure (X.509) Working Group.
206: % (PKIX, {\url{http://www.ietf.org/html.charters/pkix-charter.html}}).
207: PKIX not only profiles X.509 standards, but also
208: develops new standards apropos to the use of X.509-based PKIs in the
209: Internet.
210:
211: One of the most popular implementations of X.509-based PKI
212: is OpenSSL ({\url{http://www.openssl.org/}}, formerly SSLeay).
213: OpenSSL is a set of Open Source
214: cryptography libtaries including X.509 CA operation scripts
215: and distributed freely, such as a part
216: of PKI package for either commercial or non-commercial purpose.
217: %
218: % OpenSSL$B$OEv=i$O0E9fM"=P5,@)$N$+$+$i$J$$(BSSL$B%i%$%V%i%j$H$7$F?M5$$r(B
219: % $B=8$a!$:#F|$G$OF3F~$,4JC1$J(BPKI$B%i%$%V%i%j$H$7$F$b;H$o$l$F$$$k!%(B
220:
221: %\subsection{OpenPGP PKI}
222: %This section describes OpenPGP PKI from the historical background as well as
223: %its current use.
224:
225: % \subsubsection{PGP}
226: OpenPGP is the standard based on Pretty Good Privacy\textregistered\
227: (PGP\textregistered) application which is developed by
228: Philip Zimmermann~\cite{garfinkel94}.
229: PGP\textregistered\ is provided as
230: commercial version and `freeware' version for
231: non-commercial/non-governmental purposes only.
232:
233: The specification of PGP is standardized as OpenPGP by IETF OpenPGP Working Group.
234: %({\url{http://www.ietf.org/html.charters/openpgp-charter.html}}).
235: Today
236: ``Open\-PGP Message Format'' is defined in RFC2440~\cite{RFC2440}
237: and to be updated~\cite{RFC2440bis08}.
238: %
239: % 2002$BG/$^$G$K(BRFC$B2=$5$l$?$b$N$G!$M-8z$J$b$N$O0J2<$N$b$N$G$"$k!%(B
240: % \begin{description}
241: % \item[RFC2440, `` OpenPGP Message Format'']
242: % OpenPGP $B$rDj$a$?5,3J!%(B
243: % RFC 1991, ``PGP Message Exchange Formats''$B$O(BPGP$B%P!<%8%g%s(B2$B$r(B
244: % $B$b$H$K:n$i$l$?5,3J$@$,!$(BRFC 2440$B$O(BPGP$B%P!<%8%g%s(B5$B0J9_$r$b$H$K(B
245: % $B$7$F:n$i$l$??7$7$$5,3J$G$"$k!%(B1998$BG/H/9T!%(B\\
246: % 2003$BG/(B5$B7n8=:_!$MxMQ$G$-$k0E9f%"%k%4%j%:%`$N8+D>$7$J$I$r9T$J$C(B
247: % $B$?2~D{HG(B~\cite{RFC2440bis08}$B$r:vDjCf$G$"$k!%(B
248: % \item[RFC3156, ``MIME Security with OpenPGP'']
249: % RFC2015, ``MIME Security with Pretty Good Privacy (PGP)''$B$r2~D{$7$?$b$N!%(B
250: % $BJLL>(B OpenPGP /MIME$B$H$b8F$P$l$k!%(B2001$BG/H/9T!%(B
251: % \end{description}
252: %
253: % \subsubsection{GNU Privacy Guard}
254: The most popular OpenPGP implemenation is
255: GnuPG (GNU Privacy Guard), developed by Free Software Foundation
256: and maintained by Werner Koch of GUUG (German Unix Users Group).
257: %
258: % Version 0.0.0$B$O(B1997$BG/(B12$B7n$K%j%j!<%9$5$l!$8=:_$K;j$k$^$G3hH/$K(B
259: % $B3+H/$,?J$a$i$l$F$$$k!%(B
260: % 2002$BG/(B2$B7n$N;~E@$G:G?7HG$O(BVersion 1.0.6$B$G$"$k!%(B
261: % GnuPG$B$NFCD'$H$7$F!$(B
262: % GNU Public License$B$K=>$C$F!$>&MQHs>&MQ$rLd$o$:(B
263: % $B<+M3$K2~NI$d:FG[I[$,$G$-$k$3$H$,5s$2$i$l$k!%(B
264: % $B$^$?FC5v$G@)8B$5$l$?%"%k%4%j%:%`$rA4$/;H$C$F$$$J$$$?$a$K!$(B
265: % $BFC5v>e$NMxMQ@)8B$b$J$$$H$$$&FCD'$r;}$C$F$$$k!%(B
266: % $B$?$H$($P(BRSA$B0E9f$OFC5v@)8B$,$"$k$?$a$KI8=`%Q%C%1!<%8$K$O4^$^$l$J$+$C$?$,!$(B
267: % 2000$BG/(B9$B7n$N(BRSA$B%"%k%4%j%:%`$N(BPublic domain$B2=$K$H$b$J$$!$(B
268: % $B%P!<%8%g%s(B1.0.4$B$+$i$O(BRSA$B$,I8=`$G;H$($k$h$&$K$J$C$?!%(B
269: %
270: Either PGP or GnuPG has been known as email cryptography software firstly,
271: however they has become the general purpose data encryption tool,
272: including key exchange over Internet, trust computation, etc.
273: In the following sections, we examine the only PKI part of OpenPGP.
274: %
275: % ($B$?$@$7%$%s%?!<%M%C%H$rMxMQ$9$k%f!<%6$,:G$bB?$/MxMQ$9$k%5!<%S%9$O(B
276: % $B%a!<%k$G$"$k$N$G!$7k2LE*$K(B OpenPGP $B$,B?$/MxMQ$5$l$k>lLL$O%a!<%k$G$"$k$3(B%$B$H$O;v<B$G$"$k!%(B
277: % OpenPGP $B$O=PNO%U%)!<%^%C%H$r%P%$%J%j$b$7$/$O%"%9%-!<(B (Radix-64$B7A<0(B)$B$G(B
278: % $BNO$G$-$k$?$a!$%"%9%-!<=PNO$5$l$?7k2L$r%a!<%k$=$NB>$N%a%C%;!<%8(B
279: % $B$KD%$j$D$1$F(B
280: % $B%"%9%-!<7A<0$N%F%-%9%H%U%!%$%k$H$7$FAw$k$3$H$O%$%s%?!<%M%C%H$G$O(B
281: % $B9-$/9T$o$l$F$-$?!%(B)
282:
283: %\subsection{SPKI}
284: It is worth to point out another possible architecture, as
285: we sometimes take an closed binary question such as ``X.509 or OpenPGP,
286: which is better?'', not as
287: ``Which PKI will be the appropriate solution for different usage-scenarios?''.
288: %
289: There exists another PKI standardized by IETF --- SPKI (Simple Public Key
290: Infrastructure).
291: IETF SPKI Working Group
292: % ({\url{http://www.ietf.org/html.charters/spki-charter.html}})
293: finished its initial standardization process
294: and bring into the inter-operation stage~\cite{Ninghui2000,RFC2693}.
295: It is also called SPKI/SDSI as it is a joint force with SDSI (Simple
296: Distributed Security Infrastructure) research.
297: % by Rivest and Lampson.
298: SPKI is designed with distributed and scalable architecture in
299: many aspects, i.e., no single root CA, no globally distinguished name,
300: and flexible validity periods~\cite{Adams1991,Adam1997}.
301: %suited to the operation with WWW better than X.509
302:
303: % $B!Z:o=|![(B
304: % \begin{table}
305: % \begin{center}
306: % \begin{tabular}{|c||c|c|}
307: % \hline
308: % $BG'>ZJ}<0$N<oN`(B & $BG'>Z6I(B (CA)$B$NFC@-(B & $B<1JL;R$NFC@-(B \\
309: % \hline \hline
310: % X.509 & $B3,AX2=(B & $B%0%m!<%P%k(B\\
311: % & $BG'>Z<B;\5,Dx(B (CPS)$B$J$7(B & (X.500$B<1JL;R(B) \\
312: % \hline
313: % OpenPGP & Web of Trust & $B%0%m!<%P%k!$%f%K!<%/(B\\
314: % & & ($BEE;R%a!<%k%"%I%l%9(B)\\
315: % \hline
316: % SPKI/SDSI & $BC10l$NL?L>6I(B & $B%m!<%+%k(B \\
317: % & $BG'>Z<B;\5,Dx(B (CPS)$B$J$7(B & \\
318: % \hline
319: % SPKI without names & $B5v2DG'>Z(B & $B%0%m!<%P%k!$%f%K!<%/(B \\
320: % & $BK\?M@-$H8x3+80(BID$B$NJ,N%(B & ($B>WFM:$Fq$J8x3+80(BID)\\
321: % \hline
322: % \end{tabular}
323: % \end{center}
324: % \caption{Comparison of Certificate Types}\label{SPKI}
325: % \end{table}
326:
327: Table \ref{table:SPKI/SDSI} shows the technical comparison
328: of X.509, OpenPGP, and SPKI/SDSI
329: based on the analysis by Clarke~\cite{Clarke2001}.
330:
331: %% table $B8e$m$K0\F0(B %%
332:
333: \section{Certification without Authority}\label{web_of_trust}
334:
335: \subsection{From Face-to-Face to Web of Trust}
336: Without a certification authority,
337: the problem of trusting keys arise
338: to assess applicants before giving out certificates.
339: %
340: In OpenPGP, there are no official mechanism for creating certificates,
341: no officail channel for acquiring and distributing.
342: It makes the
343: process of certification into the face-to-face, {\it ad hoc} situation.
344: Each end user is respobsible to decide which certificate (public key of
345: an user) is trusted and accepted to be added into their local trust-file
346: (denoted ``keyring'' in PGP).
347:
348: This certification process does not require a trusted, monitored
349: registration authority or certification authority, however, it lacks
350: scalability.
351: So since PGP\textregistered\ 2.0~\cite[pp.~201--203]{Levy2001},
352: ``web of trust'' model that PGP signer acts as an introducer between
353: people had been supported.
354: % PGP 2.0 in 1992
355: % ``He[Zimmermann] didn't have the wherewithal, or money, to set up a closely
356: % monitored certification authority to distribute and verify public keys.''
357: % ~\cite[p.~201]{Levy2001}
358:
359: Figure \ref{fig:pki.eps} illustrates the model of hierarchical PKI and
360: web of trust.
361:
362: \vspace{1.0cm}
363: \begin{figure}[h]
364: \begin{center}
365: \includegraphics[width=3.0in]{pki.eps}
366: \end{center}
367: \caption{Hierarchical PKI and Web of Trust~\cite{Caronni2000}}\label{fig:pki.eps}
368: \end{figure}
369:
370: % PGP$B$O0E9f2=$7$?$j$=$l$rI|9f$9$k$@$1$G$O$J$/!$8x3+800E9f$r;H$C$?(B
371: % $BEE;R=pL>$r$9$k;v$,$G$-$k!%$7$+$7$J$,$i(BPGP$B$G$O!$(B
372: % $B$=$N8x3+80$d=pL>$,K\?M$N$b$N$G$"$k$+$I$&$+$rJ]>Z$9$k5!4X$OB8:_$;$:!$(B
373: % $BMxMQ<T$O<+J,<+?H$G?.Mj@-$rH=CG$9$k$h$&$K@_7W$5$l$F$$$k!%(B
374:
375: % OpenPGP $B$O!$(BX.509$B$GDs0F$5$l$?%b%G%k$H$O0[$J$kFH<+$N(BPKI$B$K$h$C$F(B
376: % $BEE;R=pL>$NG'>Z$r9T$J$&!%(B
377: % OpenPGP $B$G$O8x3+800E9f$,K\?M$N$b$N$G$"$k$+$I$&$+$r(B
378: % $BKvC<MxMQ<TF1;N$,8D?M%l%Y%k$GG'>Z$9$k%b%G%k$r:NMQ$7$F$$$k!%(B
379: % $B$D$^$j!$$I$N>ZL@=q$r?.Mj$7$F$I$N>ZL@=q$rGS=|$9$k$+$O(B
380: % CA$B$G$O$J$/3FMxMQ<T$,@UG$$rIi$&!%(B
381: % ``Web of Trust'' ($B?.Mj$NNX(B)$B$HI=8=$5$l$F$$$k(B
382: % PGP2.0$B$+$iDs6!$5$l$F$$$k(B~\cite[pp.~201--203]{Levy2001}$B!%(B
383:
384: % $BMxMQ<T<+?H$,H=CG$9$k:]$NH=CG:`NA$H$7$F!$(BPGP$B$O<+J,$,?.Mj$7$?>R2p<T$N(B
385: % $B=pL>$r?.MQ$9$k$H$$$&<jCJ$rDs6!$7$F$$$k!%$3$N?.Mj$N$D$J$,$j$O(B
386: % Web of Trust ($B?.Mj$NNX(B)$B$H8F$P$l$F$*$j!$(B
387: % $B$3$N?.Mj$NNX$N%M%C%H%o!<%/$,$*$h$VHO0O$K$*$$$F(B
388: % PGP$B$O%;%s%?!<$rI,MW$H$;$:$K0BA4$J%3%_%e%K%1!<%7%g%s$r9T$J$&$3$H$,$G$-$k!%(B
389:
390: \subsection{Internet Usage Scenarios}
391: OpenPGP has its own market which is different with X.509,
392: and OpenPGP community has grown in a global Internet.
393: %$B$3$l$G$OBg5,LO$JG'>Z%7%9%F%`$N1?1D$K$OIT8~$-$@$H8@$($k$+$b$7$l$J$$!%(B
394: %$B$7$+$7$J$,$i!$(BPGP/GnuPG$B$OA4@$3&5,LO$N%$%s%?!<%M%C%H%3%_%e%K%F%#$K$*$$$F(B
395: %$BMxMQ$5$l$F$-$?!%(B
396:
397: The most famous and critical use might be security alerts.
398: FIRST (Forum of Incident Response and Security Teams) and its members
399: including CERT(Computer Emergency ResponseTeam)/CC(Coordination Center)
400: have their official PGP/GnuPG public keys publicly available~\cite{First2001},
401: and have signed their alerts with their own PGP/GnuPG key.
402:
403: Usenet, operated by volunteer NetNews managers, is another example of
404: the distributed network with OpenPGP PKI. The digital signature for
405: Usenet control commands should be signed with PGP keys of represented
406: voluntary managers since 1990s~\cite{pgpcontrol}.
407: %$B$?$H$($P!$%K%e!<%9%0%k!<%W$GG[?.$5$l$k%3%s%H%m!<%k%a%C(B
408: %$B%;!<%8$X$N(BPGP$B=pL>$O(B1990$BG/Be$+$i$9$G$K<BMQ2=$5$l$F$$$k(B
409: %$B$3$l$O!$%a%C%;!<%8$rCf7Q$9$k@$3&3FCO$NL5?t$N%K%e!<%9%5!<%P$,(B
410: %$B%3%s%H%m!<%k%a%C%;!<%8(B ($B%K%e!<%9%0%k!<%W4IM}$K4X$o$kFCDj$N%a%C%;!<%8(B)$B$X$N(B
411: %$B=pL>$r<+F0E*$K8!>Z$9$k;EAH$_$G$"$k!%(B
412:
413: %$B!Z0J2<$N%U%j!<%=%U%H%&%'%"$d%3%9%H$N9M;!$O$3$3$G$O07$o$J$$!%![(B
414: %$B%K%e!<%9%0%k!<%W$X$NG[?.<u?.$K$OBh;0<T5!4X$NEPO?$OI,MW$J$/!$(B
415: %$B%K%e!<%9%5!<%P$H(BPGP/GnuPG$B$rF3F~$7$F(B
416: %$B%3%s%H%m!<%k%a%C%;!<%8MQ$N8x3+80>pJs$r@_Dj$9$l$PC/$b$,$3$NEE;R=pL>(B
417: %$B%7%9%F%`$rMxMQ2DG=$G$"$k!%(B
418: %$B$3$N%7%9%F%`$,<BMQ2=$5$l$?M}M3$H$7$F!$%=!<%9%3!<%I$*$h$S%I%-%e%a%s%H$,$9$Y$FL5(B
419: %$BNA$G8x3+$5$l$F$$$kE@$b$"$2$i$l$k!%(B
420: %INN, C News, DNews, ANU News$B$H$$$C$?<gMW%K%e!<%9%5!<%P$,(B
421: %$BB.$d$+$K$3$N5!9=$r<BAu$G$-$?$N$b!$$3$N%*!<%W%s%=!<%9$NMxE@$r@8$+$7$?(B
422: %$B$?$a$@$H9M$($i$l$k!%(B
423:
424: %$B$3$N$h$&$K8+$k$H!$(B OpenPGP $B$KE,$7$F$$$k$N$O!$(B
425: % ($B%K%e!<%9%0%k!<%W$N$h$&$K(B)
426: %$BMxMQ$K:]$7$FBh;0<T5!4X$NEPO?G'>Z$rI,MW$H$7$J$$%7%9%F%`$G$"$j!$(B
427: %$BKvC<F1;N$,?o;~%a%C%;!<%8$r8r49$7G'>Z$r9T$J$&%"%I%[%C%/$J%7%9%F%`(B
428: %(peer-to-peer$B%7%9%F%`(B)$B$G$"$k!%(B
429: %$B:#8e$5$i$J$k3+H/$*$h$S<BMQ2=$,8+9~$^$l$k(Bpeer-to-peer$B%7%9%F%`$G$O!$(B
430: %$B%;%-%e%j%F%#$d%"%+%&%s%?%S%j%F%#$rJ]>Z$9$k$?$a$K(B OpenPGP $B$N(B
431: %``Web of Trust''$B$NMxMQ$,?J$`$H9M$($i$l$k!%(B
432: %
433: %$B$^$?!$Dc%3%9%H$G(BPKI$B$r9=C[$9$k>l9g$K$b(B OpenPGP $B$*$h$S4XO"$9$k%*!<%W%s%=!<%9(B
434: %$B%=%U%H%&%'%"$NMxMQ$OM-8z$@$H9M$($i$l$k!%(B
435: %
436: %$B$=$l$KBP$7$F!$(BX.509$B$N(BPKI$B$O!$%"%I%[%C%/$JEPO?(B/$BG'>Z:n6H$K$OE,$7$F$$$J$$$,!$(B
437: %$B7QB3E*$D0BDj$7$?G'>Z%5!<%S%9$KE,$7$F$$$k!%(B
438:
439: \subsection{What is the Web of Trust?}
440: OpenPGP provides key management and certificate services
441: using local trust-file PKI.
442: The more signature is accepted, the more trust-file generated.
443: Figure \ref{fig:phillylinux.ps} is an example of a trust-file
444: visualized the relashonship of OpenPGP signature.
445: This graph illustrates who introduce the other or who meets with face-to-face,
446: in other words, whose key signed the others' public key.
447: There are no central authorities but multiplexed indivisual
448: relationships in a community.
449:
450: % $B?^(B\ref{fig:debian.ps}$B$O!$(B
451: % $B%U%j!<%=%U%H%&%'%"(BDebian$B$N3+H/<T%0%k!<%W$N3+H/<T$,;H$C$F$$$k8x3+80(B
452: % $B$N=pL>4X78$r2r@O$7!$%0%i%U2=$7$?$b$N$G$"$k!%(B
453: % 1$B8D$N8x3+80$O4v$D$+$NItIJ(B ($B%Q%1%C%H(B)$B$+$i9=@.$5$l$F$$$k!%(B
454: % $B$3$l$O8x3+80%Q%1%C%H$d8x3+80$X$N=pL>%Q%1%C%H$H8F$P$l$F$$$k$,!$(B
455: % $B$3$N%Q%1%C%H$rD4$Y$k$3$H$K$h$C$F$=$N8x3+80$K=pL>$7$F$$$k$N(B
456: % $B$,C/$J$N$+$r8!>Z$9$k$3$H$,$G$-$k!%(B
457:
458: \begin{figure}[h]
459: \begin{center}
460: \includegraphics[width=3.3in]{phillylinux.ps}
461: \end{center}
462: \caption{Visialized Web of Certificate~\cite{springgraph}}\label{fig:phillylinux.ps}
463: \end{figure}
464:
465:
466: % $B$3$N$h$&$K$7$F!$(BPGP$B$N$d$j$H$j$K$*$1$k?.MQ$NH=CG:`NA$H$J$k(B
467: % $B4X78$N$D$J$,$j!$$9$J$o$A=pL>4X78$rL\$G8+$?>l9g!$(B
468: % $B>e0L$b2<0L$b$J$$!$=E$J$j2q$&8D?MF1;N$N4X78$H$J$k$3$H$,M}2r$G$-$k!%(B
469: %
470: % $B$?$@$7!$$3$l$O?.Mj4X78$rI=$o$9$b$N$G$O$J$$!%(B
471: % $B$3$l$K$D$$$F$O<!$G@bL@$9$k!%(B
472:
473: %\subsection{Trust Signature}
474: However, this graph is not a ``web of trust'' but just a ``web of
475: certificates''.
476: OpenPGP separate the trustworthy from validity of cerfiticate.
477: For example, the amount of trust of the introducer and unknown newcomer is
478: different for an OpenPGP user.
479: Even if their certification is valid, the issuer of the key is not
480: an authority but the users own.
481: So OpenPGP users should be responsible on
482: ``Whose keys should be taken as valid but untrusted?''~\cite[p.~81]{Stajano2002}.
483:
484: In OpenPGP, users issue their signature to the other's public key with
485: their degree of trust. This is denoted ``trust signature'' and
486: represented as (trust level, trust amount)
487: with OpenPGP key management and certificate service.
488: An ordinary valid signed key is trust level 0, and
489: The signed key is asserted to be a valid trusted introducer
490: is level 1.
491: Level 2 means ``meta introducer'' or ``introducer-of-introducer''
492: that its signed key is asserted to be trusted
493: to issue level 1 trust signatures.
494: (Generally, as the introducer is more trustworthy,
495: a level $n$ trust signature asserts that a key is trusted to
496: issue level $n-1$ trust signatures.)
497: The trust amount is in a range from 0--255,
498: and appointed 60 for partial trust and 120 for complete trust~\cite[s.v. ``Trust Signature'']{RFC2440}.
499: As OpenPGP distinguished the trust from validity,
500: ``web of trust'' is also distinguished from ``web of certificates''.
501:
502: % ``introducer'' is not ``certification authority''
503:
504: \subsection{PGP Revocation Problem}\label{revocation}
505:
506: The weakest link of OpenPGP PKI is the revocation of public
507: key~\cite[pp.~585--586]{Schneier1995}~\cite[p.~309]{P2P2001:ch16}.
508: As there is no official channel for acquiring and distributing OpenPGP
509: public keys, there are no guarantee about
510: how to tell everyone that your key is no longer valid.
511:
512: % OpenPGP $B$rB>$N(BPKI$B$HHf3S$7$?:]$K!$KvC<MxMQ<T$,DL?.Aj<j$N80$r$I$N$h$&$K(B
513: % $BF~<j$7?.Mj$9$k$N$+$H$$$&LdBj$r@h$K=R$Y$?!%$=$7$F$5$i$K?<9o$JLdBjE@$b(B
514: % $BB8:_$9$k!%$=$l$O!$8x3+80$N<:8z$N=hM}$G$"$k!%(B
515: % $B$?$H$($PO31L$d2~cb$K$h$C$F<+J,$,G[$C$?8x3+80$,;H$($J$/$J$C$?;~!$(B
516: % $B0J2<$NLdBjE@$,;XE&$5$l$F$$$k(B
517: % ~\cite[pp.~585--586]{Schneier1995}~\cite[p.~309]{P2P2001:ch16}
518: % \begin{itemize}
519: % \item $BHkL)80$r$J$/$7$F$7$^$C$?$i8x3+80$r<h$j>C$9$3$H$,$G$-$J$$(B
520: % \item $B8x3+80$r<h$j>C$7$F$b!$8x3+80$rG[I[$7$?A40w$K<h$j>C$7>pJs$,EA$o$kJ]>Z$,$J$$(B
521: % \end{itemize}
522: % $B$3$l$O$$$:$l$b(BPGP$B$N1?MQ$K$*$1$k8x3+80$N<h$j>C$7%7%9%F%`$,IT==J,$G$"$k(B
523: % $B$?$a$K5/$3$kLdBj$G$"$k!%(B
524: %
525: % ``The weakest link of this[PGP] whole system is key revocation: It is
526: % impossible to guarantee that no one will use a compromised key. If
527: % Alice's private key is stolen she can send out something called a {\bf
528: % key revocation certificate}, but since key distribution is {\it ad hoc}
529: % and largely word of mouth there is no guarantee that it will reach
530: % everyone who has her public key on his key ring. And as Alice has to
531: % sign the key revocation certificate with her private key; if she loses
532: % the key altogether she cannot revoke it.'' ~
533: % ~\cite[pp.~585--586]{Schneier1995}
534: %
535: % ``How do you tell everyone that your key is no longer valid? How do you
536: % tell everyone that your certificate on a key should be changed? For that
537: % matter, what exactly did Bob mean when he certified Charlie's key, and
538: % does Charlie mean the same thing when he certifies David's key?''
539: % ~\cite[p.~309]{P2P2001:ch16}
540:
541: % $B8x3+80$NM-8z4|4VCf$KFbMFJQ99!&HkL)80$NEpFq!&J6<:!&GK2u$J$I$,@8$8$?$?$a$K(B
542: % $B8x3+80>ZL@=q$,<:8z$7$?>l9g!$MxMQ<T$KDLCN$9$kI,MW$,@8$8$k!%(B
543: % $B$3$N>l9g!$(BX.509$B$G$O(BCA$B$,$=$l$>$l$N(B
544: % $B1?1DJ}?K$K1~$8$F>ZL@=q<:8z%j%9%H(B (CRL)$B$r:n@.$7!$O"7H$9$k(BCA$B$X$HG[I[$9$k!%(B
545: % $B$=$l$KBP$7$F(B OpenPGP $B$G$O!$MxMQ<T<+?H$,GK4~>ZL@=q$r:n@.$7!$(B
546: % $B3FMxMQ<T$XG[I[$9$k!%$=$N:]$NLdBjE@$H$7$F!$(B
547: % $BMxMQ<T$X$NDLCN$,:$Fq$J$3$H!$$5$i$KMxMQ<T<+?H$,HkL)80$rJ6<:$7$?:]$K$O(B
548: % $BGK4~>ZL@=q$9$i:n@.$G$-$J$/$J$k!%(B
549: % $B$3$NLdBj$KBP$7$FK\O@J8$G$O80%5!<%P$r;H$&$3$H$,M-8z$@$H9M$($k!%(B
550: % $B$3$N8!F$$K$D$$$F$O80%5!<%P$N@a$G=R$Y$k!%(B
551:
552: % PGP certificates are revoked when private keys are compromised (or users
553: % forget the passwords locking their private keys). In X.509, only the
554: % certificate's issuer can revoke a certificate. In PGP, "only the
555: % certificate's owner (the holder of its corresponding private key) or
556: % someone whom the certificate's owner has designated as a revoker can
557: % revoke a PGP certificate. (Designating a revoker is an useful practice,
558: % as it's often the loss of the passphrase for the certificate's
559: % corresponding private key that leads a PGP user to revoke his or her
560: % certificate - a task that is only possible if one has access to the
561: % private key.)"14
562: % As PGP does not use commercial CAs, the PKI does not
563: % use CRLs for revocation.
564: % ~\cite[pp.~56--57]{Clarke2001}
565:
566: The typical answer to this revocation problem of PGP is to use
567: PGP public keyserver for distributing certification.
568: ``Typically, to communicate that a certificate has been revoked, a signed
569: note, called a key revocation certificate, is posted on PGP certificate
570: servers, and widely distributed to people who have the key on their
571: public keyrings. People wishing to communicate with the affected user,
572: or use the affected key to authenticate other keys, are warned about the
573: hazards of using that public key''~\cite[pp.~56--57]{Clarke2001}.
574: However, there are few research on the PGP public keyserver and
575: usually the keyserver is not considered as the part of OpenPGP PKI.
576: In the following, this paper examines PGP keyserver
577: as the part of OpenPGP PKI.
578:
579: \section{Related Works}\label{sec:related_works}
580: There are several research fields related to OpenPGP public keyserver.
581: The first is the study on the traditional PGP public keyservers,
582: %(the most of keyserver is still based on PGP\textregistered\
583: %application and not fully supports OpenPGP standard)
584: the second is the integrated channel for OpenPGP key distribution,
585: and the third is the combined ``web of trust'' with other PKI.
586:
587: %\subsection{Keyserver-related study}
588: A ``web of trust'' used in PGP is referred in several
589: researches including the peer-to-peer authentication~\cite{P2P2001:ch16},
590: trust computation~\cite{Maurer96b,Caronni2000},
591: and privacy enhanced technology~\cite{garfinkel94}.
592: % and textbooks~\cite{Schneier1995,Anderson2001}.
593: However, there are few description on PGP keyserver.
594: It might be because PGP keyserver mechanism is too simple.
595: It is not a CA but just a pool of public keys.
596: % In the next section we propose the life cycle of a digital certificate
597: % with PGP keyserver testing the next generation OpenPGP keyserver.
598: From users' viewpoint, PGP keyserver has a
599: large amount of OpenPGP public keys that provide
600: the interesting material for social analysis of network community.
601: For example, OpenPGP keyserver developer Jonathan McDowell
602: also developed ``Experimental PGP key path finder''~\cite{McDowell2002}
603: that searches and displays the chain of certification between the users.
604: % This ``six degrees of separation'' effect~\cite[p.~309]{P2P2001:ch16}
605:
606: %\subsection{Providing a root of a global community}\label{Register}
607: As OpenPGP's initial trust file is blank, the users have to start with a
608: face-to-face certificate to exchange public keys.
609: Though another initial file is provided via high integrity channel.
610: {\it Global Internet Trust Register}~\cite{Register}
611: is a printed book that contains ``fingerprints'' (hash values of certificate)
612: of the most important public keys (mainly cryptography experts who are
613: likely to have signed many other keys in their respectice
614: communities)~\cite[pp.~80--81]{Stajano2002}.
615:
616: %\subsection{Combined OpenPGP PKI}
617: OpenPGP PKI itself can be described as the superset of PKI~\cite{Zimmermann2001},
618: however, combining OpenPGP PKI with other authentication system is
619: challenging work in both theoretical and operational field.
620: %
621: Formal study of trust relationship of PKI started in the late 1990s~\cite{Maurer96b,Caronni2000} and GnuPG development version in December 2002
622: started to support its trust calculation with
623: GnuPGP's trust signature.
624:
625: The implementation of trust calculation is ongoing and
626: using large-scale ``web of trust'' (not ``web of certificates'')
627: is not so popular outside of computer experts.
628: On the other hands, using different types of PKI has become more popular.
629:
630: In the early work at MIT,
631: PGP-signing service had been combined with Kerberos CA
632: system that does not have public key cryptography~\cite{Schiller1995}.
633: % In this scheme, the authenticity of the certificate's name-to-key
634: % binding is as sound as the Kerberos account's creation was. If the
635: % user-accounts administrator checked IDs in face-to-face meetings, the
636: % Kerberized CA's certificates will be meaningful. If instead the users
637: % can register themselves remotely, then the certificates will be all but
638: % meaningless. (At MIT, staff members usually get their accounts
639: % face-to-face)
640: Today, the hybrid system of OpenPGP and X.509 is
641: both developed into some OpenPGP implementations.
642: In 2001, German authorities
643: BSI (Bundesamt f\"{u}r Sicherheit in der Informationstechnik,
644: Germany's agency for information technology security)
645: accept the \"{A}gypten project for Open Source implementation
646: of governmental mail user agents Sphinx
647: which supports X.509v3, PKCS, LDAP, and OpenPGP ~\cite{Newsforge20011005}.
648: The results of the open development are begun to
649: imported to other commercial products in 2002--2003~\cite{Kmail}.
650: In a same way, PGP Corporation also released PGP\textregistered\ version
651: 8.0 as X.509-enabled application that can interoperate X.509
652: certificates and CAs~\cite{PGP8X509}.
653:
654:
655: % $B$^$?%I%$%D7P:QO+F/>J(B (BMWA)$B$OFbL3>J$H6&F1$GEE;R>&<h0z$HEE;R@/I\$r?d?J$9$k(B
656: % GNU Privacy Project$B$r?d?J$7$F$*$j!$9qFb@=IJ$N3+H/$@$1$G$J$/(B
657: % $BF|K\H/$N%a!<%k%/%i%$%"%s%H(BSylpheed$B$N%I%$%D8lHG$NG[I[$b9T$J$C$F$$$k(B
658: % $B$[$+!$(B
659: % $B30It6H<T$K0QBw$7$F(B
660: % {\it GnuPP for beginners},
661: % {\it GnuPP for experts}
662: % $B$H$$$C$?(B
663: % $B%I%$%D8l$N<j0z=q$b@):n!&G[I[$r9T$J$C$F$$$k(B~\cite{GnuPP}$B!%(B
664: % $B$3$N<j0z=q$O%3%_%C%/Iw$NA^3($D$-$G8x3+800E9f$N;EAH$_$d(B
665: % $BEE;R>ZL@=q$N(Bfingerprint$B$N3NG'J}K!!$(B
666: % $B$=$7$F8x3+80%5!<%P$N@$3&E*$J%M%C%H%o!<%/$H$$$C$?(B
667: % $B6qBNE*$JFbMF$r07$C$F$$$k!%(B
668: % (GnuPP / WinPP Project)
669:
670: % $B>ZL@=qG[I[$NLdBj$O(B OpenPGP $B$N$_$J$i$:(BX.509$B$K$*$$$F$O$5$i$K(B
671: % $BCm0U$,I,MW$G$"$k$3$H$r9M$($?>l9g(B~\cite{Takagi2002:CSS}$B!$(B
672: % $B<j0z=q$G(Bfingerprint$B$N3NG'$r$H$j$"$2$k$3$H$O9b$/I>2A$G$-$k!%(B
673: % **
674:
675: \section{OpenPGP Public Keyserver}\label{sec:keyserver}
676: Before describing our research, this section describes
677: OpenPGP keyserver generally.
678: Keyserver is not a CA. It only pool anyone's public keys.
679: Keyserver never issue any certificate for someone's public key
680: but only provide it.
681:
682: \subsection{Current Status}
683: The first keyserver is developed at MIT in 1994
684: by Brian~ A.~LaMacchia. It exchange public keys with email
685: and keys are managemented with PGP command.
686: For users, keyserver acts as an easy email agent
687: who receives any valid but untrusted keys,
688: then searches and provides the key to everyone.
689: %
690: % (Software Design, 2002$BG/(B8$B7n9f(B)
691: %
692: % $B$3$l$O(B\ref{sec:PKI}$B@a$G=R$Y$?(Bstart point $B$H$7$F$N(B trust-file$B$r(B
693: % $B8x3+$7$F;2>H$9$k$b$N$@$H8@$&$3$H$,$G$-$k(B~\cite[ch.~15]{garfinkel94}
694: %
695: In 1997, PGP Public KeyServer (pksd,
696: {\url{http://www.mit.edu/people/marc/pks/}}) started
697: by MIT student Marc Horowitz.
698: Pksd uses a database management system and has been working fine.
699: The database system is operated via email, CGI-interface from http server,
700: and HKP --- pksd's own communication protocol over
701: Hypertext Transfer Protocol (HTTP).
702: In 2003, David Shaw of GnuPG team proposed the OpenPGP HTTP Keyserver
703: Protocol~\cite{HKP200303} based on traditional HKP
704: as the draft for Internet Standard.
705:
706: Today pksd has been working fine even if in global distributed
707: network. There are 10 or more syncronized public keyservers in the world
708: and the most of them are running with patched pksd.
709: These public keyservers are operated by voluntary managers belong to
710: organizations including
711: \special{html:<a href ="http://pgp.mit.edu/">}
712: MIT
713: %{\url{http://pgp.mit.edu/}})
714: \special{html:</a>}
715: and
716: \special{html:<a href ="http://www.cc.gatech.edu/~peter/pks-commands.html">}
717: Georgia Tech
718: %({\url{http://www.cc.gatech.edu/~peter/pks-commands.html}})
719: \special{html:</a>}
720: in United States,
721: \special{html:<a href ="http://pki.surfnet.nl/extract.html">}
722: SURFnet
723: %({\url{http://pki.surfnet.nl/extract.html}})
724: \special{html:</a>}
725: in Netherlands,
726: \special{html:<a href ="http://www.dfn-pca.de/pgpkserv/">}
727: DFN-CERT
728: \special{html:</a>}
729: %({\url{http://www.dfn-pca.de/pgpkserv/}})
730: in Germany,
731: \special{html:<a href ="http://www.rediris.es/cert/servicios/keyserver/">}
732: RedIRIS (IRIS-CERT)
733: %({\url{http://www.rediris.es/cert/servicios/keyserver/}})
734: \special{html:</a>}
735: in Spain,
736: \special{html:<a href ="http://pgp.nic.ad.jp/">}
737: JPNIC
738: % ({\url{http://pgp.nic.ad.jp/}})
739: \special{html:</a>}
740: in Japan.
741: % Horowitz's pksd-0.9.4 + patch level 2
742: % $B>&MQ$@$H(B Veridis will continue the development and full support of
743: % Highware OpenKeyServer product and the KeyServer.Net service.
744: % $B$"$k$$$O(BCERT$B$,1?1D$7$F$$$k!%(B)
745: Today they have more than 1,400,000 public keys entries and
746: 3,000/day or more transactions between each sync sites.
747: In 2000,
748: SURFnet held the first PGP keyserver manager symposium~\cite{SURFnet2000}
749: and the managers keep in touch with each other.
750:
751: \subsection{Revocation process and Keyserver}
752: As public keyservers provides semi-official key distribution channel,
753: keyserver adds powerful feature to OpenPGP PKI.
754: % $BMxMQ<T8D?M$,G'>Z$K@UG$$r;}$D(B OpenPGP $B$N(BPKI$B$O!$80%5!<%P$N5!G=$rAH$_9g$o$;$k(B
755: % $B$3$H$K$h$C$F$5$i$KM-8z$J(BPKI$B$H$J$k$H8@$($k!%(B
756: %
757: Public keyservers can handle the PGP revocation problem that we described in
758: section \ref{revocation}.
759: % ``The weakest link of this[PGP] whole system is key revocation: It is
760: % impossible to guarantee that no one will use a compromised key. If
761: % Alice's private key is stolen she can send out something called a {\bf
762: % key revocation certificate}, but since key distribution is {\it ad hoc}
763: % and largely word of mouth there is no guarantee that it will reach
764: % everyone who has her public key on his key ring. And as Alice has to
765: % sign the key revocation certificate with her private key; if she loses
766: % the key altogether she cannot revoke it.''
767: % ~\cite[pp.~585--586]{Schneier1995}
768: % $BHkL)80$r$J$/$7$F$7$^$C$?$i8x3+80$r<h$j>C$9$3$H$,$G$-$J$$(B
769: % $B8x3+80$r<h$j>C$7$F$b!$8x3+80$rG[I[$7$?A40w$K<h$j>C$7>pJs$,EA$o$kJ]>Z$,$J$$(B
770: %
771: % $BA0<T$N<eE@$K4X$7$F$O!$<+J,$N8x3+80$HHkL)80$r:n@.$9$k:]$K(B
772: % $B8x3+80$N<h$j>C$7$K;H$&GK4~>ZL@=q$r:n@.$7$F(B
773: % $BHkL)80$H$OJL$N>l=j$KJ]B8$9$k$H$$$C$?<j=g$rF'$^$($?1?MQ(B
774: % $B$K$h$C$F$"$kDxEY2sHr$9$k;v$,2DG=$G$"$k!%(B
775: %
776: % ``How do you tell everyone that your key is no longer valid? How do you
777: % tell everyone that your certificate on a key should be changed? For that
778: % matter, what exactly did Bob mean when he certified Charlie's key, and
779: % does Charlie mean the same thing when he certifies David's key?''
780: % ~\cite[p.~309]{P2P2001:ch16}
781: %
782: Using keyserver provides an answer to the question
783: ``How do you tell everyone that your key is no longer valid?''.
784: User may issue a suicide note (denoted as ``revocation signature'' in OpenPGP )
785: and post it to keyservers.
786: Receiving a valid revocation signature, keyserver updates the key to be
787: revoked. The update key with revocation signature is redistributed to
788: the synchronized keyservers in the world,
789: and finally PGP user updates their keyrings with the nearest keyserver.
790: The updated key with valid revoked signature makes users's older key not
791: to be used.
792:
793: % $B8e<T$NLdBj$K$D$$$F$O!$8x3+80%5!<%P$N3hMQ$,M-8z$G$"$k!%<+J,$N8x3+80$rF~<j(B
794: % $B$7$??M$9$Y$F$K8x3+80$N99?7>pJs$rEA$($k$N$OIT2DG=$K6a$$!%$@$,!$8x3+80%5!<(B
795: % $B%P$r2p$;$P99?7<j=g$rBgI}$K>JNO2=$9$k$3$H$,$G$-$k!%8x3+80%5!<%P$K8x3+80$N(B
796: % $B99?7>pJs(B ($B$^$?$OGK4~>ZL@(B)$B$rAw$l$P!$%5!<%P>e$KEPO?$5$l$F$$$k8x3+80$K99?7(B
797: % $B>pJs$,2C$($i$l!$85$N8x3+80$r;}$C$F$$$k?M$b8x3+80%5!<%P$KLd$$9g$o$;$l$P$$(B
798: % $B$D$G$b<+J,$N;}$C$F$$$k8x3+80$,99?7$5$l$F$$$J$$$+$I$&$+$r3NG'$9$k$3$H$,$G(B
799: % $B$-$k!%(B
800:
801: %\subsection{``Don't believe keyserver too much''}
802: %$BB>?M$N(BPGP$B8x3+80$r4JC1$KF~<j$G$-$k$3$H$,(B
803: %$B8x3+80%5!<%P$NMxE@$@$H$h$/8@$o$l$F$$$k!%(B
804: %$B$7$+$7(BPGP$B$N8x3+80$O<+M3$JL>A0$G:n@.$9$k$3$H$,$G$-$k$N$G!$(B
805: %$B8x3+80%5!<%P$KEPO?$5$l$F$$$k8x3+80$,K\Ev$K%f!<%6$,:n$C$?(B
806: %$B$b$N$+$I$&$+$O$o$+$i$J$$!%(B
807: %$B$D$^$j!$8x3+80%5!<%P$GF~<j$7$?8x3+80$r;H$&$K$O(B
808: %$B$^$:8x3+80$,56B$$5$l$F$$$J$$K\J*$G$"$k$3$H$r3NG'$9$kI,MW$,$"$k!%(B
809: %$B$3$NJ}K!$H$7$F$O!$(BPGP$B%/%i%$%"%s%H$,Aj<j$N8x3+80$+$i<h$j=P$9(B
810: %$B%O%C%7%eCM!V%U%#%s%,!<%W%j%s%H!W$r3NG'$9$k$N$,0lHLE*$G$"$k!%(B
811:
812: \subsection{Current Keyserver Problem}
813: Today's sisutation around PGP keyserver
814: is beyond the original developers' idea, and
815: current pksd also has some limitations.
816: %neither an expandable format by IETF standard
817: %nor the distributed network of public keyservers.
818:
819: Firstly, the implementations of pksd are not OpenPGP-compliant.
820: OpenPGP ~\cite{RFC2440} published in 1998
821: defines two versions of signature formats.
822: (Version 3 provides basic OpenPGP signature information, while version 4 provides
823: an expandable format with subpackets.)
824: These changes made traditional PGP applications
825: not-OpenPGP-compliant --- not only
826: PGP\textregistered~\cite[s.v. ``Implementation Nits'']{RFC2440}
827: but also pksd.
828: Today pksd does not fully support OpenPGP format.
829:
830: Seconary, the pksd does not scalability for global use.
831: Though pksd has simple but strong dabatabase management system,
832: it is neither soshisticated nor scalable compared with
833: today's Internet server.
834: For the matter, pksd cannot handle 1 billion keys and cannot
835: accept such many transactions as Yahoo! or eBay site.
836: % Like as ihave/sendme, mirror DB or something good
837: New design of OpenPGP public keyserver is required.
838:
839: \section{OpenPKSD: Next Generation OpenPGP Public
840: Keyserver}\label{OpenPKSD}
841: We introduce our next generation OpenPGP Public Keyserver project
842: with a new architecture. We call it OpenPKSD
843: ( OpenPGP Public KeyServer Daemon).
844: It is developed by one of the
845: authors and funded by Japanese Information-technology Promotion Agency
846: (IPA) in 2001--2002.
847:
848: \subsection{Server Design and Implementation}
849: OpenPKSD supports OpenPGP subpacket format
850: and works as high-performance server with SQL backend.
851: The design of OpenPKSD oriented to not only high-performance,
852: but also flexible extension capability and easy operation.
853: We implemented OpenPKSD with Ruby and
854: PostgreSQL backend~\cite{RubyConf2002}.
855:
856: \subsection{User Interface and Security}
857: As ``Web of trust'' depends on users' decision,
858: user interface is also important factor on security.
859: For example, Whitten and Tygar~\cite{Whitten1999} had
860: ever pointed out some dangerous errors occured with past PGP clients'
861: interface.
862:
863: Users can connect to OpenPKSD with two kind of
864: interfaces, OpenPGP client or CGI on WWW.
865: Providing WWW interface, OpenPKSD
866: must help users' recognization, judgement, and
867: handling on OpenPGP public keys.
868:
869: %\paragraph{No fingerprint display}
870:
871: %$B$=$N$?$a$K$O8x3+80%5!<%P$NI=<($r?.Mj$9$k$@$1$G$OIT==J,$G!$(B
872: %$B<+J,$N%m!<%+%k%^%7%s>e$N(BPGP$B%/%i%$%"%s%H$G(B
873: %$B$=$N8x3+80$,?.MQ$G$-$k$+$r3NG'$9$Y$-$G$"$k!%(B
874: %$B$3$NJ}K!$H$7$F$O!$(BPGP$B%/%i%$%"%s%H$,Aj<j$N8x3+80$+$i<h$j=P$9(B
875: %$B!V%U%#%s%,!<%W%j%s%H!W$r3NG'$9$k$N$,0lHLE*$G$"$k!%(B
876:
877: OpenPKSD displays only 64bit KeyID to identify someone's public keys.
878: Though some other servers calculate and display ``fingerprint'' of
879: public keys before download it, it does not help users arare of risk
880: using keyserver.
881: As keyserver is just a pool and not CA,
882: users should check the public key with their own.
883: Moreover, it is easy to make some faked keyserver by
884: Man-in-the-Middle Attack, TCP hijacking, etc.
885: It means that
886: the fingerprint must be calculated under user's (safe) machine
887: and that is the reason why OpenPKSD does not display fingerprint.
888:
889: %\paragraph{Interface with pgpdump: visualizing subpackets}
890: OpenPKSD WWW interface provides additional feature
891: to visualize subpackets of PGP keys.
892: As OpenPKSD has an expandable format with subpackets,
893: it is very hard to understand the data structure inside this.
894: Using pgpdump program,
895: key packet visualizer that displays the packet format of
896: OpenPGP and PGP\textregistered\ version 2.
897:
898: Many PGP users are familiar with this verification
899: on added keys, as in 2000, PGP\textregistered\ version 5.5.x to 6.5.3 had
900: a serious security hole that cannot detected with fingerprint
901: verification.
902: Then CERT/CC had alarted
903: ``Check certificates for ADKs [Additional Decryption Keys]
904: before adding them to a keyring.''~\cite{CERT:CA-2000-18}
905: Pgpdump exactly visualizes these additional keys.
906: With pgpdump, OpenPKSD helps users to recognize the information of
907: public key and any other added keys before downloading.
908:
909: \subsection{Performance and Future work}
910: OpenPKSD is implemented with Ruby language and PostgreSQL DBMS.
911: Ruby is so-called ``scripting language'' and seemed not suitable
912: for a quick response or large program development.
913: However,
914: OpenPKSD succeeds not only the more compact code size but also
915: quick response compared with pksd,
916: by loading bit calculation modules such as CRC24 checksum
917: written by C language~\cite{Hironobu2003}.
918: Table \ref{table:OpenPKSD} shows the performance of
919: OpenPKSD version 0.2.8, non-cluster version, installed on PC.
920: %%table
921: \begin{table}[htbp]
922: \begin{center}
923: \begin{tabular}{|c|c|}
924: \hline
925: CPU: & Intel Pentium4 1.6GHz \\
926: \hline
927: HDD: & IDE ATA100 7200rpm 60GB\\
928: \hline
929: Memory: & PC2100 768MB \\
930: \hline
931: One key query: & 120ms average,\\
932: & 72ms best,\\
933: & 230ms worst.\\
934: \hline
935: \end{tabular}
936: \caption{OpenPKSD Performance}\label{table:OpenPKSD}
937: \end{center}
938: \end{table}
939: % [pgp-folks]
940: % OpenPKSD (v0.2.8) Performance:
941: %
942: % CPU: Intel P 4 1.6Ghz
943: % HDD: IDE ATA100 7200rpm
944: % MEM: PC2100 768MB
945: %
946: % One key query: 120ms Ave., 72ms best, 230ms worst.
947: % If a public key has 10 signatures, I estimate total time of queries
948: % become around 700ms - 2000ms. I'm afraid that it makes not only
949: % machine heavy load situation but also become target of DoS attack.
950: %
951: OpenPKSD version 0.2.8 is also working
952: well at handling usual transaction between other PGP public keyserver
953: described in section \ref{sec:keyserver} since 2002.
954:
955: Forthcoming developers' version of OpenPKSD will support some clustering
956: based on the reserch on the performance of
957: cluster technology~\cite{Wang2002:ICPADS}.
958: It will be published in 2003 and
959: support the experimental HKP(keyserver protocol over http) balancer,
960: keyserver cluster, and clustered database.
961:
962: \section{Summary}
963: In this paper, we overlooked some PKI architectures.
964: %$BK\>O$G$O!$(BPKI$B$N;kE@$+$i(BX.509$B$*$h$S(B OpenPGP $B$r(B
965: %$B$H$j$"$2!$N><T$,0[$J$k%b%G%k$N<B8=$rL\;X$7$F$$$k$3$H!$$*$h$S(B
966: %$BN><T$NFCD9$r<($7$?!%$^$?!$$=$l0J30$N(BPKI$B$NDs0F$H$7$F(B
967: %SPKI/SDSI$B$K$D$$$F$b>R2p$7$?!%(B
968: %PKI$B$K$OM#0l$N%b%G%k$OB8:_$7$J$$!%$7$?$,$C$F(B
969: %PKI$B$r9=C[$9$k$K$"$?$C$F$I$N%b%G%k$r:NMQ$9$k$N$+$H$$$&;kE@$,I,MW$H$J$k!%(B
970: %$B$9$J$o$A!$N><T$OAj8_GS=|$N4X78$K$"$k$N$G$O$J$/!$J;B82DG=$G$"$k!%(B
971: %$B:#8e$O!$I,MW$H$5$l$k>u67$K1~$8$F(B
972: %$BAPJ}$NMxE@$r@8$+$7$?%$%s%?!<%M%C%H(BPKI$B$N9=C[$rL\;X$9$Y$-$G$"$k!%(B
973: %
974: Using ``Web of Trust,'' OpenPGP PKI can help users to
975: manage certification without CAs.
976: However, there are the problem on public key management, i.e.,
977: how to get the receivers' public key, or,
978: how to tell everyone that the public key is no longer valid.
979: PGP keyserver is the solution to the problem.
980: %PKI$B$H$7$F$N(B OpenPGP $B$O!$G'>Z6I5Z$SEPO?5!4X$r2p$9$k$3$H$J$/(Bpeer-to-peer$B$NG'>Z$r(B
981: %$B9T$J$($k$H$$$&FCD9$,$"$k!%$7$+$7$J$,$i!$(B
982: %$B!VDL?.$7$?$$Aj<j$N8x3+80$r$I$&$d$C$F<j$K$$$l$k$+!W(B
983: %$B$=$7$F!V8x3+80$NGK4~>ZL@$r$I$&$d$C$FDLCN$9$k$+!W$H$$$&E@$,IT==J,$G$"$k!%(B
984: %$B$3$NLdBjE@$X$N<h$jAH$_$H$7$F!$80%5!<%P$NLr3d$r$"$2$?!%(B
985:
986: Though some PGP public keyservers have built a global PKI,
987: traditional PGP keservers have some limitations.
988: We introduced OpenPKSD,
989: newly-designed and OpenPGP-supported
990: public keyserver project.
991: OpenPKSD took its first step, works well in practice, and
992: examining the cluster technology.
993:
994: \section*{Availability}
995: OpenPKSD source code and documents
996: are available under GNU General Public License (GPL) at
997: {\url{http://www.openpksd.org/}}.
998:
999: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Bibliography
1000: %\bibliographystyle{latex8}
1001: %\bibliography{research,yamane}
1002:
1003: \begin{thebibliography}{10}\setlength{\itemsep}{-1ex}\small
1004:
1005: \bibitem{Adams1991}
1006: C.~Adams and S.~Lloyd.
1007: \newblock {\em Understanding Public-Key Infrastructure: Concepts, Standards,
1008: and Deployment Considerations}.
1009: \newblock Macmillan Technical Publishing, 1999.
1010:
1011: \bibitem{Register}
1012: R.~J. Anderson, B.~Crispo, J.-H. Lee, C.~Manifavas, and V.~{Maty\'{a}\v{s}
1013: Jr.}, editors.
1014: \newblock {\em Global Internet Trust Register}.
1015: \newblock MIT Press, 1999 edition, Mar. 1999.
1016:
1017: \bibitem{RFC2440}
1018: J.~Callas, L.~Donnerhacke, H.~Finney, and R.~Thayer.
1019: \newblock {OpenPGP Message Format}.
1020: \newblock {{\it {R}equest {F}or {C}omments}}, Nov. 1998.
1021: \newblock RFC 2440 (Category: Standards Track) replaces RFC 1991, ``PGP Message
1022: Exchange Formats''.
1023:
1024: \bibitem{RFC2440bis08}
1025: J.~Callas, L.~Donnerhacke, H.~Finney, and R.~Thayer.
1026: \newblock {OpenPGP Message Format}.
1027: \newblock {{\it {Internet Draft}}}, May 2003.
1028: \newblock Revision of RFC 2440\cite{RFC2440}.
1029: {\url{http://www.ietf.org/internet-drafts/draft-ietf-openpgp-rfc2440bis-08.txt}} (visited June 1, 2003).
1030:
1031: \bibitem{Caronni2000}
1032: G.~Caronni.
1033: \newblock Walking the web of trust.
1034: \newblock In {\em Proceedings of IEEE 9th International Workshops on Enabling
1035: Technologies: Infrastructure for Collaborative Enterprises (WET ICE'00)},
1036: pages 153--158. IEEE Computer Press, 2000.
1037:
1038: \bibitem{CERT:CA-2000-18}
1039: {CERT/CC}.
1040: \newblock {PGP May Encrypt Data With Unauthorized ADKs}.
1041: \newblock CERT Advisory CA-2000-18, 2000.
1042: \newblock {\url{http://www.cert.org/advisories/CA-2000-18.html}} (visited May
1043: 31, 2003).
1044:
1045: \bibitem{Clarke2001}
1046: D.~E. Clarke.
1047: \newblock {SPKI/SDSI HTTP Server} / {Certificate Chain Discovery in SPKI/SDSI}.
1048: \newblock Master's thesis, MIT, Sept. 2001.
1049: \newblock {\url{http://theory.lcs.mit.edu/~cis/theses/clarke-masters.pdf}}
1050: (visited May 27, 2003).
1051:
1052: \bibitem{springgraph}
1053: Darxus@ChaosReigns.com.
1054: \newblock springgraph.
1055: \newblock Online document, May 2002.
1056: \newblock {\url{http://www.chaosreigns.com/code/springgraph/}} (visited May 27,
1057: 2003).
1058:
1059: \bibitem{P2P2001:ch16}
1060: R.~Dingledine, M.~J. Freedman, and D.~Molnar.
1061: \newblock Accountability.
1062: \newblock In A.~Oram, editor, {\em Peer-to-Peer: Harnessing the Power of
1063: Disruptive Technologies}, chapter~16. O'Reilly \& Associates, Mar. 2001.
1064:
1065: \bibitem{RFC2693}
1066: C.~M. Ellison, B.~Frantz, B.~Lampson, R.~Rivest, B.~Thomas, and T.~Yl\"{o}nen.
1067: \newblock {SPKI Certificate Theory}.
1068: \newblock {{\it {R}equest {F}or {C}omments}}, Sept. 1999.
1069: \newblock RFC 2693 (Status: Experimental).
1070:
1071: \bibitem{First2001}
1072: FIRST.ORG.
1073: \newblock {{FIRST} Contact Information}.
1074: \newblock Online document, 2001.
1075: \newblock {\url{http://www.first.org/docs/contact.html}} (visited May 31,
1076: 2003).
1077:
1078: \bibitem{garfinkel94}
1079: S.~Garfinkel.
1080: \newblock {\em {PGP}: Pretty Good Privacy}.
1081: \newblock O'Reilly \& Associates, Dec. 1994.
1082:
1083: \bibitem{Adam1997}
1084: R.~Khare and A.~Rifkin.
1085: \newblock Weaving a web of trust.
1086: \newblock {\em {World Wide Web Journal}}, 2(3), Summer 1997.
1087: \newblock Online article available at
1088: {\url{http://www.w3journal.com/7/s3.rifkin.wrap.html}} (visited May 29,
1089: 2003).
1090:
1091: \bibitem{pgpcontrol}
1092: D.~C. Lawrence.
1093: \newblock {Authentication of Usenet Group Changes}.
1094: \newblock Online document, 1999.
1095: \newblock {\url{ftp://ftp.isc.org/pub/pgpcontrol/README.html}} (visited May 27,
1096: 2003).
1097:
1098: \bibitem{Levy2001}
1099: S.~Levy.
1100: \newblock {\em Crypto: How the Code Rebels Beat the Government, Saving Privacy
1101: in the Digital Age}.
1102: \newblock Viking Penguin, Jan. 2001.
1103:
1104: \bibitem{Ninghui2000}
1105: N.~Li.
1106: \newblock Local names in {SPKI/SDSI}.
1107: \newblock In {\em Proceedings of 13th IEEE Computer Security Foundations
1108: Workshop (CSFW-13)}, pages 2--15. IEEE Computer Society, 2000.
1109:
1110: \bibitem{Maurer96b}
1111: U.~Maurer.
1112: \newblock Modeling a public-key infrastructure.
1113: \newblock In E.~Bertino, editor, {\em Proceedings of 1996 European Symposium on
1114: Research in Computer Security ({ESORICS}' 96)}, Lecture Notes in Computer
1115: Science 1146, pages 325--350. Springer-Verlag, 1996.
1116:
1117: \bibitem{McDowell2002}
1118: J.~McDowell.
1119: \newblock Experimental {PGP} key path finder.
1120: \newblock Online document, 2002.
1121: \newblock {\url{http://the.earth.li/~noodles/pathfind.html}} (visited May 31,
1122: 2003).
1123:
1124: \bibitem{Newsforge20011005}
1125: {KMail} and mutt as {Sphinx}-clients for {German} authorities.
1126: \newblock {\em NewsForge}, 5 October 2001.
1127: \newblock Online article available at
1128: {\url{http://www.newsforge.com/article.pl?sid=01/10/05/162238}} (visited May
1129: 27, 2003).
1130:
1131: \bibitem{PGP8X509}
1132: {PGP Corporation}.
1133: \newblock Using an {X.509 PKI} with {PGP}\textregistered\ 8.0: Protecting
1134: existing investments.
1135: \newblock {PGP Coporation} {White Paper}, May 2003.
1136: \newblock {\url{http://www.pgp.com/products/whitepapers/PGP8X509.pdf}} (visited
1137: July 1, 2003).
1138:
1139: \bibitem{Schiller1995}
1140: J.~I. Schiller and D.~Atkins.
1141: \newblock Scaling the web of trust: Combining {Kerberos} and {PGP} to provide
1142: large scale authentication.
1143: \newblock In {\em Proceedings of {USENIX} 1995 Technical Conference on UNIX and
1144: Advanced Computing Systems}, pages 83--94. {USENIX} Association, 1995.
1145:
1146: \bibitem{Schneier1995}
1147: B.~Schneier.
1148: \newblock {\em Applied Cryptography: Protocols, Algorithms, and Source Code in
1149: C}.
1150: \newblock John Wiley \& Sons, 2nd edition, 1995.
1151:
1152: \bibitem{Kmail}
1153: S.~Shankland.
1154: \newblock {KDE Linux} adds {German} flavor.
1155: \newblock {\em ZDNet News}, 31 January 2003.
1156: \newblock Online article available at
1157: {\url{http://zdnet.com.com/2100-1104-982816.html}} (visited May 27, 2003).
1158:
1159: \bibitem{HKP200303}
1160: D.~Shaw.
1161: \newblock {OpenPGP HTTP Keyserver Protocol (HKP)}.
1162: \newblock {{\it {Internet Draft}}}, Mar. 2003.
1163: \newblock
1164: {\url{http://www.ietf.org/internet-drafts/draft-shaw-openpgp-hkp-00.txt}}
1165: (visited May 28, 2003).
1166:
1167: \bibitem{FYI36}
1168: R.~W. Shirey.
1169: \newblock {Internet Security Glossary}.
1170: \newblock {{\it {R}equest {F}or {C}omments}}, May 2000.
1171: \newblock RFC 2828 (Also FYI 36) (Status: Informational).
1172:
1173: \bibitem{Stajano2002}
1174: F.~Stajano.
1175: \newblock {\em Security for Ubiquitous Computing}.
1176: \newblock John Wiley \& Sons, Feb. 2002.
1177:
1178: \bibitem{SURFnet2000}
1179: {SURFnet}.
1180: \newblock {The First PGP Keyserver Manager Symposium}.
1181: \newblock Online document, 2000.
1182: \newblock {\url{http://www.surfnet.nl/bijeenkomsten/pgp/}} (visited May 27,
1183: 2003).
1184:
1185: \bibitem{RubyConf2002}
1186: H.~Suzuki.
1187: \newblock {OpenPKSD: OpenPGP Public Keyserver}.
1188: \newblock Presentation slide, 2002.
1189: \newblock {\url{http://openpksd.org/docs/RubyConf2002.pdf}} (visited May 28,
1190: 2003). Presented at 2nd Annual International Ruby Conference, held in
1191: Nov.~1--3, 2002, Seattle, WA.
1192:
1193: \bibitem{Hironobu2003}
1194: H.~Suzuki.
1195: \newblock {OpenPKSD: Next Generation OpenPGP Public Keyserver} implementation
1196: with {Ruby} (in {Japanese}).
1197: \newblock In {\em Proceedings of Software Symposium 2003}, pages 162, 220--223,
1198: Japan, July 2003. Software Engineers Association.
1199:
1200: \bibitem{Wang2002:ICPADS}
1201: J.~Wang, Y.~Tsutaya, N.~Segawa, S.~Yamane, Y.~Murayama, M.~Miyazaki, and
1202: H.~Suzuki.
1203: \newblock Approaches to balancing data load of shared-nothing clusters and
1204: their performance comparison.
1205: \newblock In {\em Proceedings of 9th International Conference on Parallel and
1206: Distributed Systems (ICPADS 2002)}, pages 293--301. IEEE Press, Dec. 2002.
1207:
1208: \bibitem{Whitten1999}
1209: A.~Whitten and D.~Tygar.
1210: \newblock Why {Johnny} can't encrypt: A usability evaluation of {PGP 5.0}.
1211: \newblock In {\em Proceedings of the 9th USENIX Security Symposium}, Aug. 1999.
1212: \newblock Also available at
1213: {\url{http://www.usenix.org/publications/library/proceedings/sec99/full_papers/whitten/whitten.ps}}, more detailed presentation is available at
1214: {\url{http://reports-archive.adm.cs.cmu.edu/anon/1998/abstracts/98-155.html}}
1215: (visited May 31, 2003).
1216:
1217: \bibitem{Zimmermann2001}
1218: P.~Zimmermann.
1219: \newblock {Why OpenPGP's PKI is better than an X.509 PKI}.
1220: \newblock Online document, 27 Feb 2001.
1221: \newblock {\url{http://www.openpgp.org/technical/whybetter.shtml}} (visited May
1222: 27, 2003).
1223:
1224: \end{thebibliography}
1225:
1226:
1227: \onecolumn
1228: \begin{table}[htbp]
1229: \begin{center}
1230: {\small
1231: \begin{tabular}{|l|r|l|}
1232: \hline
1233: {\bf X.509} & CA Characteristics: & Global Hierarchy. There are commercial X.509 CAs. \\
1234: & & X.509 communities are built from the top-down. \\
1235: % & CA $B$NFCD'(B: & $B%0%m!<%P%k$J3,AX!%%3%^!<%7%c%k$N(BX.509 CA$B$bB8:_$9$k(B\\
1236: % & & X.509$B$N%3%_%e%K%F%#$O%H%C%W%@%&%s$K$h$C$F9=C[$5$l$k!%(B\\
1237: & Trust Model:
1238: & Hierarchical Trust Model. Trust originates from a `trusted' \\
1239: & & CA, over which the guardian may or may not have control. \\
1240: & & A requestor provides a chain of authentication from the \\
1241: & & `trusted' CA to the requestor's key. \\
1242: % & $B?.MQ%b%G%k(B:
1243: % & $B3,AX7?$N?.MQ%b%G%k!%?.MQ$O!V?.MQ$G$-$k!W(BCA$B$,$D$/$k!%(B\\
1244: & Signatures: & Each certificate has one signature, belonging to the issuer \\
1245: & & of the certificate. \\
1246: % & $BEE;R=pL>(B: & $B$=$l$>$l$N>ZL@=q$O>ZL@=qH/9T<T$K$h$C$FEE;R=pL>$5$l$F$$$k!%(B\\
1247: & Certificate Revocation: & Uses CRL(Certificate Revocation List)s \\
1248: % & $B>ZL@=q$N<:8z(B: & CA$B$,(BCRL (Certificate Revocation List)$B$rG[I[$9$k!%(B \\
1249: & Name Space: & Global \\
1250: % & $B%M!<%`%9%Z!<%9(B: & $B%0%m!<%P%k(B \\
1251: & Types of Certificates: & Name Certificates \\
1252: % & $B>ZL@%?%$%W(B & $BK\?M@-$N>ZL@(B \\
1253: & Name-to-Key binding:
1254: & Single-valued function: each global name is bound to ex- \\
1255: & & actly one key (assuming each user has a single public- \\
1256: & & private key pair). \\
1257: % & $B<1JLL>$H80$H$N7k9g(B:
1258: % & $BC10lCM4X?t(B: $B%0%m!<%P%k$J<1JLL>$O$=$l$>$lM#0l$N(B \\
1259: % & & $B80%Z%"$H0lBP0lBP1~$7$F$$$k!%(B\\
1260: % & & ($B%f!<%6$O$?$@$R$H$D$N8x3+80HkL)80%Z%"$r;}$C$F$$$k$H2>Dj$9$k(B) \\
1261: \hline
1262: {\bf OpenPGP} & CA Characteristics: & Egalitarian design. Each key can issue certificates. \\
1263: \quad {\bf } & & PGP communities are built from the bottom-up in a\\
1264: & & distributed manner. \\
1265: % & CA $B$NFCD'(B: & $BJ?Ey<g5AE*$J%G%6%$%s!%$=$l$>$l$N80$G>ZL@=q$rH/9T$G$-$k!%(B\\
1266: % & & PGP$B$N%3%_%e%K%F%#$OJ,;69=B$$N%\%H%`%"%C%W$K$h$C$F9=C[$5$l(B\\
1267: % & & $B$k!%(B\\
1268: & Trust Model: & {\it Web of Trust}, file-based PKI.\\
1269: % & $B?.MQ%b%G%k(B: & {\it Web of Trust} ($B?.MQ$NNX(B) \\
1270: & Signatures: & Each certificate can have multiple signatures; the first \\
1271: & & signature belongs to the issuer of the certificate. \\
1272: % & $BEE;R=pL>(B: & $B$=$l$>$l$N>ZL@=q$OJ#?t$N=pL>$r4^$`$3$H$,$G$-$k!%:G=i$N(B
1273: % $B=pL>$O(B\\
1274: % & & $B>ZL@=qH/9T<T$K$h$k=pL>$G$"$k!%(B\\
1275: & Certificate Revocation: & A `key revocation certificate,' suicide note is posted on \\
1276: & & {\it PGP keyservers}, and widely distributed to people who\\
1277: & & have the compromised key on their public keyrings. \\
1278: % & $B>ZL@=q$N<:8z(B & $BMxMQ<T$,GK4~>ZL@=q$r(BPGP$B8x3+80%5!<%P$KEPO?$9$k$3$H$G!$(B\\
1279: % & & $BL58z$K$J$C$?8x3+80$r<j85$K;}$C$F$$$kAj<j$KG[I[$9$k!%(B\\
1280: & Name Space: & Global \\
1281: % & $B%M!<%`%9%Z!<%9(B: & $B%0%m!<%P%k(B \\
1282: & Types of Certificates: & Name Certificates \\
1283: % & $BG'>Z%?%$%W(B: & $BK\?M@-$NG'>Z(B \\
1284: & Name-to-Key binding:
1285: & Single-valued function: each global name is bound to ex- \\
1286: & & actly one key (assuming each user has a single public-\\
1287: & & private key pair). \\
1288: % & $B<1JLL>$H80$H$N7k9g(B:
1289: % & $BC10lCM4X?t(B: $B%0%m!<%P%k$J<1JLL>$O$=$l$>$lM#0l$N(B\\
1290: % & & $B80%Z%"$H0lBP0lBP1~$7$F$$$k!%(B\\
1291: % & & ($B%f!<%6$O$?$@$R$H$D$N8x3+80HkL)80%Z%"$r;}$C$F$$$k$H2>Dj$9$k(B) \\
1292: \hline
1293: {\bf SPKI/~SDSI} & CA Characteristics: & Egalitarian design. The principals are the public keys. \\
1294: {\bf } & & Each key can issue certificates. SPKI/SDSI communities \\
1295: & & are built from the bottom-up in a distributed manner. \\
1296: % & CA$B$NFCD'(B: & $BJ?Ey<g5AE*$J%G%6%$%s!%$=$l$>$l$N80$O>ZL@=q$rH/9T$G$-$k!%(B\\
1297: % & & $B8x3+80$,<1JL@-$r;}$D!%(B\\
1298: % & & SPKI/SDSI$B$N%3%_%e%K%F%#$OJ,;69=B$$N%\%H%`%"%C%W$K$h$C$F9=(B\\
1299: % & & $BC[$5$l$k!%(B\\
1300: & Trust Model: & Trust originates from the guardian. A requestor provides \\
1301: & & a chain of authorization from the guardian to the \\
1302: & & requestor's key. The infrastructure has a clean, scalable \\
1303: & & model for defining groups and delegating authority. \\
1304: % & $B?.MQ%b%G%k(B: & {\it chain of authorization}$B$NDs6!(B \\
1305: % & & $B%0%k!<%W$rDj5A$7G'>Z$r0QG$$9$k$?$a$K%$%s%U%i%9%H%i%/%A%c$O(B\\
1306: % & &$BL@2w$G%9%1!<%i%V%k$J%b%G%k$r;}$D!%(B\\
1307: & Signatures: & Each certificate has one signature, belonging to the
1308: issuer\\
1309: & & of the certificate.\\
1310: % & $BEE;R=pL>(B: & $B$=$l$>$l$N>ZL@=q$O>ZL@=q$NH/9T<T$K$h$k=pL>$r0l$D4^$s$G$$(B
1311: % $B$k!%(B\\
1312: & Certificate Revocation: & Advocates using short validity periods
1313: and {\it Certificates of}\\
1314: & & {\it Health}.\\
1315: % & $B>ZL@=q$N<:8z(B: & $BC;$$M-8z4|8B$H@8B8>ZL@=q$rMQ$$$k(B\\
1316: & Name Space: & Local \\
1317: % & $B%M!<%`%9%Z!<%9(B: & $B%m!<%+%k(B \\
1318: & Types of Certificates: & Name Certificates,
1319: Authorization Certificates \\
1320: % & $B>ZL@%?%$%W(B: & $BK\?M@-$N>ZL@$^$?$OG'>Z$N>ZL@(B \\
1321: & Name-to-Key binding: & Multi-valued function: each local name is bound to zero, \\
1322: & & one or more keys (assuming each user has a single public \\
1323: & & -private key pair). \\
1324: % & $B<1JLL>$H80$H$N7k9g(B:
1325: % & $BB?CM4X?t(B: $B%m!<%+%k$JL>A0$O%<%m!$(B1$B!$$=$l0J>e$N80$H7k9g$5$l$k!%(B\\
1326: % & & ($B%f!<%6$O$?$@$R$H$D$N8x3+80HkL)80%Z%"$r;}$C$F$$$k$H2>Dj$9$k(B) \\
1327: \hline
1328: \end{tabular}}
1329: \end{center}
1330: \caption{Comparison of X.509, OpenPGP, and SPKI/SDSI}\label{table:SPKI/SDSI}
1331: \end{table}
1332: \twocolumn
1333:
1334: %\section{memo}
1335: %http://packages.debian.org/stable/misc/signing-party.html
1336:
1337: \end{document}
1338: %
1339: %\appendix
1340: %\section{Brainstorming}
1341: %
1342: %\paragraph{Untrusted CA}
1343: % $B>&MQ$NG'>Z6I$O3N<B$+(B?
1344: %
1345: % $B$+$D$F%^%$%/%m%=%U%H<R$K@.$j$9$^$7$?2?<T$+$,!"(B
1346: % Verisign$B<RH/9T$NG'>Z=q$r<hF@$7A{$.$K$J$C$?!%(B
1347: % $B$3$N;v7o$,H/3P$9$k$H(BVerisign$B$O(B
1348: % $BH/9T$7$?$3$NIT@5$JG'>Z=q$rL58z$K$7!$$=$l$rDLCN$9$k(BCRL
1349: % (Certificate Revocation List)$B$rMQ0U$7$?!%(B
1350: % $B$7$+$7!$Ev$N%^%$%/%m%=%U%H$N@=IJ$K$O(BVeriSign$B$,MQ0U$7$?(BCRL$B$r=hM}$9$k5!(B
1351: % $BG=$,<BAu$5$l$F$$$J$+$C$?$?$a!$(BOS$B$r%"%C%W%G!<%H$9$kI,MW$K$J$C$?(B
1352: % ~\cite{MS01-07}$B!%(B
1353: %
1354: % OpenPGP $B$N$h$&$K<+J,$N<j$G3NG'$9$kJ}K!$+!"$"$k$$$O(BWeb of Trust$B$N$h$&$K(B
1355: % $B$"$J$?$,?.Mj$G$-$kBh;0<T$N>ZL@$r$"$J$?$,3NG'$9$kJ}K!$O$H$F$b9gM}E*$J$N(B
1356: % $B$G$9!#Bh0l$K@UG$=j:_$,L@3N$G$"$k$3$H$,>e$2$i$l$^$9!#$3$l$OMxMQ<T@UG$$G(B
1357: % $B$"$k$3$H$,$O$C$-$j$7$F$$$^$9!#<!$K?.Mj$N$*$1$kJ#?t$NBh;0<T$N=pL>$,$"$k(B
1358: % $B$h$&$J>l9g!"$9$Y$F$N=pL>$,56B$!"$"$k$$$O@.$j$9$^$7$G$@$^$;$k3NN($O>.$5(B
1359: % $B$/$J$j$^$9!#(B
1360:
1361: %\subsection{ OpenPGP $B$H(BX.509$B$H$NHf3S(B}
1362: %Table \ref{table:PGP/X509}$B$K(BPKI$B$H$7$F$N(B OpenPGP $B$H(BX.509$B$H$rHf3S$9$k!%(B
1363: %\begin{table}
1364: % \begin{center}
1365: % \begin{tabular}{|r|l|l|}
1366: % \hline
1367: % ~ & X.509 & OpenPGP \\ \hline
1368: % PKI architecture: & hierarchical PKI & trust-file PKI \\
1369: % Certificate Issuer : & CA & each user \\
1370: % Base of trust: & root CA & Face-to-face \\
1371: % $BG'>Z$NO":?9=B$(B: & $B%D%j!<7?(B & $B%f!<%6Cf?47?(B \\
1372: % $BG'>Z<T$rG'>Z$9$k:,5r(B: & $BMxMQ<T$K$h$kA*Br(B & $BMxMQ<T<+?H(B \\
1373: % $B>ZL@=q$NGK4~$N4IM}(B: & $B$"$j(B & $BIT40A4(B\\
1374: % $B%3%9%H(B: & $B9b$$(B & $BDc$$(B \\
1375: % \hline
1376: % \end{tabular}
1377: % \end{center}
1378: % \caption{X.509$B$H(B OpenPGP $B$NAj0cE@(B}\label{table:PGP/X509}
1379: %\end{table}
1380:
1381: %\subsection{$B8x3+80%5!<%P$NMxMQ=87W;vNc(B}
1382: %$B8x3+80%5!<%P$KEPO?$5$l$F$$$kKDBg$J8x3+80%G!<%?$rE}7WE*$K(B
1383: %$B2r@O$9$k;n$_$b9T$J$o$l$F$$$k!%(B
1384: %1996$BG/(B1$B7n$K(BNeal McBurnett$B$,:G=i$NE}7W$r9T$J$$!$(B
1385: %$B8=:_$G$O(BDrew Streib$B$,(B
1386: %$B9q:]E*$JF14|$r$H$C$F$$$k(BPGP$B8x3+80%5!<%P$N0l$D$G$"$k(B {\tt www.us.pgp.net
1387: % (pgp.dtype.org)}$B$N%G!<%?$rDj4|E*$KJ,@O$7$F$$$k!%(B
1388: %
1389: %$BC1$J$k8x3+80?t$N=87W0J30$K$b!$(B
1390: %$B6=L#?<$$;n$_$H$7$F!$(B
1391: %$B%5!<%P$KEPO?$5$l$F$$$k8x3+80$+$i(B
1392: %$BC/$,C/$K=pL>$7$F$$$k$+$r=87W$7!$=pL>?t$N>e0L<T$rD4$Y$kD4::$,$"$k!%(B
1393: %$B$3$ND4::$K$h$l$P!$%8%g!<%8%"9)2JBg3X$N(B Peter N. Wan $B$H(BMIT$B$N(B Theodore Ts'o $B$,(B
1394: %$B<s0L$rAh$C$F$$$k!%(B
1395: %
1396: %% [pgp-folks.124]
1397: %% $B80%5!<%P$KEPO?$5$l$F$$$k8x3+80$N=pL>4X78$r2r@O$9$k3X@8$H$$$&$N$O!$(B
1398: %% Neal McBurnett $B$+(B Drew Streib $B$N$3$H$G$9$M!%(B
1399: %% http://dtype.org/keyanalyze/
1400: %% $B$+$i!$$?$H$($P5nG/$N(B8$B7n$N2r@O7k2L$r8+$F$_$k$H!$$d$O$j(B1$B0L$O(BPeter Wan.
1401: %% http://dtype.org/keyanalyze/200108.php
1402:
1403: %\subsection{$B%-!<%5%$%s%Q!<%F%#!<(B}
1404: %%$BH`$i$N$h$&$KL5?t$N?MJ*$N8x3+80$K=pL>$r9T$J$&$?$a$K$O!$(B
1405: %%$BB?$/$N?M$,8x3+80$r;}$C$F8=$l$?;~$K!$$=$N?MJ*$,(B
1406: %%$B$?$7$+$K$=$N8x3+80$N;}$A<g$G$"$k$3$H$r(B
1407: %%$B3NG'$9$k<j=g$r$"$i$+$8$a7h$a$F$*$/I,MW$,$"$k!%(B
1408: %
1409: %% CA$B$G$O$J$/(BBOF
1410: %% ~\cite{Rose199804}
1411: %
1412: %$B8x3+80$X$N=pL>$r9T$J$&A0$K9T$J$&$3$N<j=g$H$7$F$O!$(B
1413: %$BB?$/$N?M$r=8$a$F9T$J$&%-!<%5%$%s%Q!<%F%#!<$,$h$/CN$i$l$F$$$k!%(B
1414: %
1415: %$B%-!<%5%$%s%Q!<%F%#!<$K$O!$(B
1416: %$B;22C<T$,0lF2$K=8$^$j!$(B
1417: %$BA40w$,4i$r8+9g$o$;$J$,$i%Q%9%]!<%H$r2sMw$9$k$J$I$7$F(B
1418: %$BAj8_$KG'>Z$9$kJ}K!!$$"$k$$$O(B
1419: %$BG'>ZC4Ev<T$,K\?M3NG'$7!$$=$N>l$K5o9g$o$;$?%a%s%P!<$,(B
1420: %$B$=$3$G3NG'$5$l$?8x3+80$K=pL>$r9T$&$H$$$&(B2$B$D$N<jK!$,$"$k!%(B
1421: %
1422: %$BA0<T$O(B OpenPGP $B$N%a%C%7%e7?G'>Z%Q%9$r3hMQ$7$F$*$j!$(B
1423: %$B$*8_$$$rCN$j?FKS$r?<$a$k$H$$$&8z2L$b4|BT$G$-$k$,!$(B
1424: %$B?M?t$,A}$($l$PA}$($k$[$I8zN($,0-$/$J$k$?$a$K(B
1425: %$B?M?t$N5,LO$,>.$5$$>l9g$K8B$i$l$k!%(B
1426: %
1427: %% $BA0<T$O%P%C%I%"%$%G%"$@!%$3$l$O?M$,A}$($k$K$D$lHs>o$KLLE]$K$J$k$N$OL\$K(B
1428: %% $B8+$($F$$$k$7!$<B:]$K$d$C$F$_$k$HHa;4$G$"$k!%(B
1429: %% 2000$BG/(B4$B7n$K(BGNUPG$B$N(BWerner
1430: %% Koch$B!$(BPGP$B$N(BPhilip Zimmermann$B$r$O$8$aI.<T$b4^$a$?(BPGP keyserver manager
1431: %% 20$B?tL>$,0lF2$K=8$^$C$?>.$5$$$,Hs>o$K=EMW$J2q5D$,%*%i%s%@$G3+:E$5$l$?!%(B
1432: %% $B$3$N;~$H$P$+$j$KA40w$GA40w$rG'>Z$9$k%?%$%W$N%-!<%5%$%s%Q!<%F%#!<$r9T$C(B
1433: %% $B$?!%%Q%9%]!<%H$J$I$N<L??F~$j?HJ,>ZL@=q$,A40w$N4V$G2sMw$5$l!$<+J,$N=gHV(B
1434: %% $B$K$J$C$?;~$K<+J,$G<+J,$N%U%#%s%,!<%W%j%s%H$rFI$_>e$2$k!%$*8_$$$rCN$j?F(B
1435: %% $BKS$r?<$a$k$H$$$&8z2L$O@dBg$G$"$k$,Hs>o$K8zN($,0-$/BgJQ$@$C$?!%(B
1436: %
1437: %$BAH?%Fb$NMxMQ$H$7$F$O!$8e<T$NJ}K!$,0lHLE*$G$"$k!%6qBNE*$K$O!$(B
1438: %$B%;%-%e%j%F%#C4Ev<T$K;vA0$K8x3+80$H%U%#%s%,!<%W%j%s%H$rAw$j!$(B
1439: %$BLL@\$N8e$KC4Ev<T$,8x3+80$KEE;R=pL>$r;\$9$H$$$C$?J}K!$G$"$k!%(B
1440: %$B$?$H$($P%Z%s%7%k%t%!%K%"Bg3X$G$O!$%;%-%e%j%F%#C4Ev<T$K(BPGP$B=pL>(B
1441: %$B$r?=@A$9$k<jB3$-$d%,%$%I%i%$%s$r8x3+$7$F$$$k(B~\cite{upenn}$B!%(B
1442: %
1443: %$B%*!<%W%s%=!<%93+H/<T$N4V$G$b!$%5%$%s%Q!<%F%#!<$N<j0z$-(B
1444: %$B$,G[I[$5$l$F$$$k(B~\cite{keyparty}
1445: %$B$^$?!$%-!<%5%$%s%Q!<%F%#!<$N3+:EM=9p$r%$%s%?!<%M%C%H>e$G8x3+$9$k(B
1446: %$B%5%$%H$bB8:_$9$k(B~\cite{partyguide}$B!%(B
1447: %$B9q:]AH?%$H$7$F$O(B
1448: %IETF$B$NG/(B4$B2s$NBg2q$K$*$$$F$b!$(B
1449: %IETF Night Session $B$G$N(B PGP Key Signing
1450: %$B$,3+$+$l$k$N$,91Nc$K$J$C$F$*$j!$(BPGP$B%5%$%s%Q!<%F%#!<$O0lDj$N(B
1451: %$B5!G=$r2L$?$7$F$-$?$H8@$($k!%(B
1452: %
1453: %
1454: %% Linkname: Slides for the Usenix 1995 Keynote Address by Mark Weiser,
1455: %% 9am CST January 18, 1995.
1456: %% URL: http://www.ubiq.com/hypertext/weiser/Usenix95.html
1457: %
1458: %\paragraph{Practical Viewpoint}
1459: %% Hacker **
1460: %% Taylorist productivity measures (FirstMonday)
1461:
1462: % \subsection{German Government and GnuPG}
1463: % 1999$BG/0J9_!$%I%$%D@/I\$O(BGnuPG$B%W%m%8%'%/%H$r;Y1g$7!$@/I\5!4X$K:NMQ$9$k(B
1464: % $BF0$-$r8+$;$F$$$k!%(B
1465: % $B$^$:(B1999$BG/(B11$B7n$K(B
1466: % $B%I%$%D$NO"K.7P:Q5;=Q>J(B (BMWi, Bundesministerium f\"{u}r Wirtschaft und
1467: % Technologie)$B$,(BGnuPG$B%W%m%8%'%/%H$K(B318,000$B%^%k%/(B ($B$*$h$=(B17$BK|(BUS$B%I%k(B)$B$N(B
1468: % $B;q6b1g=u$r9T$J$&$3$H$rH/I=$7$?(B~\cite{BMWi1999,Wayner1999}$B!%(B
1469: % $B$3$N%W%m%8%'%/%H%j!<%@!<$O%I%$%D:_=;$N(BWerner Koch$B$G$"$k(B
1470: % (GnuPG$B$O@$3&3FCO$+$i3+H/<T$,;22C$7$F$$$k$,!$%W%m%8%'%/%H%j!<%@!<$O(B
1471: % Werner Koch$B$,C4Ev$7$F$$$k(B)$B!%(B
1472: %
1473: % GnuPG$B$N=i4|%P!<%8%g%s$N%j%j!<%9$HA08e$7$F%I%$%D@/I\$O(B
1474: % $B%=!<%9%3!<%I$r8x3+$K$9$k$3$H$G%7%9%F%`$N0BA4@-$r3NJ]$9$k$?$a$K!$(B
1475: % $B0E9f$NG[I[(B/$BM"=P$KBP$7$F@)8B$r@_$1$J$$J}?K$rH/I=$7$F$$$k$,!$(B
1476: % GnuPG$B$N;Y1g$O$=$N1dD9$H$7$F9M$($i$l$k!%(B
1477:
1478: %\paragraph{visualization}
1479: % \begin{figure}[h]
1480: % \begin{center}
1481: % \includegraphics[width=6.0in]{debian-keyring.dot.1024.ps}
1482: % \end{center}
1483: % \caption{Visualized Web of Certificate~\cite{springgraph}}\label{fig:debian.ps}
1484: % \end{figure}
1485:
1486: % $BBP>]$H$J$C$?(BDebian$B3+H/%A!<%`%a%s%P!<$O!$(B
1487: % $B%$%s%?!<%M%C%H>e$G$N3+H/$N$d$j$H$j$K$O(B OpenPGP $B$K=`5r$7$?(B
1488: % GNU Privacy Guard (GnuPG, GPG)$B$rMQ$$$?=pL>$r(B
1489: % $B8x<0$KMQ$$$F$$$k!%(B
1490: % $B$3$N;k3P2=$9$k$3$H$K$h$C$F!$(B
1491: % $B%W%m%8%'%/%H%j!<%@!<$,Cf?4E*$JLr3d$r(B
1492: % $B2L$?$7$F$$$k$3$H$,3NG'$5$l$?(B~\cite{Debian2000}$B!%(B
1493:
1494: %\subsection{Characteristics of OpenPGP }
1495: % ``X.509 vs PGP'' $BBPN)35G0$G9M$($F$7$^$$$,$A$G$"$k!%(B
1496: % ``Why OpenPGP's PKI is better than an X.509 PKI''~\cite{Zimmerman2001}
1497:
1498: % X.509$B$H(B OpenPGP $B$O$?$H$(F1$80E9f%"%k%4%j%:%`$r:NMQ$7$F$$$F$b!$(B
1499: % $B$3$l$^$G$^$C$?$/JL!9$KI8=`2=$r?J$a$F$*$j!$Aj8_@\B3@-$O$J$+$C$?!%(B
1500: % $B$?$H$($P!$(BS/MIME$B=`5r$NEE;R=pL>%a!<%k$H(B OpenPGP $B=`5r$NEE;R=pL>%a!<%k$H$O(B
1501: % $BJL!9$N%"%W%j%1!<%7%g%s$K$h$C$F=hM}$5$l$k!%(B
1502: % $B$3$l$ON><T$N(BPKI$B$N%b%G%k$,0[$J$C$F$$$k$?$a$G$"$k!%(B
1503:
1504: % $B$3$N$?$a$K!$(B OpenPGP $B$ODL?.$7$?$$Aj<j$N8x3+80$r(BCA$B$+$iF~<j$9$k$N$G$O$J$/!$(B
1505: % $BKvC<MxMQ<T$,<+NO$GF~<j$7$J$1$l$P$J$i$J$$!%(B
1506: % $B$3$l$OG'>Z%7%9%F%`9=C[$K$*$$$FLdBj$H$J$k2DG=@-$,$"$k!%(B
1507:
1508: % 1997$BG/$ND4::$K$h$l$P!$<B:]$K;H$o$l$F$$$k(B OpenPGP $B$N(Bweb of trust$B$G$O(B
1509: % 6$B$D@h$^$G$NG'>Z4X78$^$G$7$+$?$I$l$J$+$C$?(B~\cite{McBurnett1997}$B!%(B
1510:
1511: % Paper Submission: July 1, 2003
1512: % Author Notification: Sep. 15, 2003
1513: % Final Manuscript: Oct. 7, 2003
1514: %http://www.saint2004.org
1515: %
1516: % Please note that a paper should be strictly limited within 10 pages,
1517: % and extra charge will be applied for pages 8 to 10 when accepted for
1518: % the proceedings.
1519: % - 1 to 7 page : no extra charge
1520: % - 8 to 10 page : 21000 YEN (about $175) per page
1521: % - more than 10 page : rejected
1522: % Deadline: July 1, 2003, 23:59:59 Apia, Samoa time
1523: %
1524: % ispell -W 9 -V -t main.tex
1525: %dvips -f main.dvi | ghostview -
1526: