bundle.xml

这是一个C的源代码

<chapter>
<title>Bundled Programs</title>
<para>
Several binaries and curve parameters are bundled with the PBC library,
such as the <command>pbc</command> program
<link linkend="pbcintro">discussed earlier</link>.
</para>
<para>
The <filename>param</filename>
subdirectory contains pairing parameters one might use in
a real cryptosystem. Many of the test programs read the parameters
from files such as these on standard input, for example:
</para>
<screen>benchmark/benchmark &lt; param/c159.param
example/bls &lt; param/e.param
</screen>
<section id="sect.pbc">
<title>Pairing-Based Calculator</title>
<para>
The <filename>pbc</filename> subdirectory contains a program also named
<filename>pbc</filename> which allows interactive testing of the PBC library.
</para>
<para>
Its syntax is loosely based on that of <filename>bc</filename>, a well-known
arbitrary precision calculator language. I do not intend for
<filename>pbc</filename> to be a fully-fledged scripting language. If this
were the aim I would add a module to an extant popular language.
Instead, pbc is meant to be a test program.
(Nonetheless, it possesses enough functionality to
be used to implement cryptosystems in shell scripts.)
</para>
<para>
Four pairings named A, D, E and F are initialized during startup, and
furthermore, the variables G1, G2, GT and Zr are setup to represent groups
associated with the A pairing. To use a different pairing, call the init_pairing
funciton.
</para>
<para>
Assignments have a C-like syntax "variable = expression", and return the
value of the variable.
Function calls and expressions also resemble C.
The arithmetic operators +, -, /, *, ^ have the standard precedence.
Perhaps the only significant difference is that no terminating semicolon
is required.
</para>
<para>
Comments begin with "#" and end at a newline.
Double quotes are used to denote strings.
On error, pbc prints the error code which I have yet to translate to English.
</para>
<para>
Some of the pbc functions:
</para>

<variablelist>
<varlistentry><term>
<command>init_pairing</command>(<parameter>pairing</parameter>)
</term>
<listitem><para>
Set the variables G1, G2, GT and Zr to the groups in
<parameter>pairing</parameter>, e.g:
</para>
<screen>
init_pairing(D)
</screen>
</listitem>
</varlistentry>

<varlistentry><term>
<command>rnd</command>(<parameter>G</parameter>)
</term>
<listitem><para>
Returns a random element of the group G, e.g:
</para>
<screen>
g = rnd(Zr)
</screen>
<para>
Has synonyms <command>rand</command> and <command>random</command>.
</para></listitem>
</varlistentry>

<varlistentry><term>
<command>pairing</command>(<parameter>g</parameter>, <parameter>h</parameter>)
</term>
<listitem><para>
Returns the pairing applied to
<parameter>g</parameter> and <parameter>h</parameter>.
The element <parameter>g</parameter> must be an element of G1
and <parameter>h</parameter> of G2, e.g:
</para>
<screen>
pairing(rnd(G1), rnd(G2))
</screen>
</listitem>
</varlistentry>

<varlistentry><term>
<command>fromstr</command>(<parameter>string</parameter>, <parameter>G</parameter>)
</term>
<listitem><para>
Returns <parameter>string</parameter> converted to an element of
<parameter>G</parameter>, e.g.
</para>
<screen>
x = fromstr("[123,456]", GT)
</screen>
</listitem>
</varlistentry>

</variablelist>
</section>

<section>
<title>Parameter Generation</title>
<para>
Programs that generate pairing parameters are located in the
<filename>gen</filename> subdirectory. Some of the programs are already
functional enough to be used to find parameters for real applications.
I need to write more documentation first; for now, read
the source!
</para>
<variablelist>
<varlistentry>
<term>listmnt</term>
<listitem><para>
Searches for discriminants D that lead to MNT curves with subgroups
of prime order.
</para></listitem>
</varlistentry>
<varlistentry>
<term>genaparam</term>
<term>gena1param</term>
<term>gendparam</term>
<term>geneparam</term>
<term>genfparam</term>
<listitem><para>
Prints parameters for a curve suitable for computing pairings of a given
type. The output can be fed to some of the other test programs.
</para>
<para>
The program <command>gendparam</command>
should be given a discriminant as the first argument.
</para></listitem>
</varlistentry>
<varlistentry>
<term>hilbertpoly</term>
<listitem><para>
Prints the Hilbert polynomial for a given discriminant (the first argument).
Computing the Hilbert polynomial is an intermediate step when generating
type D parameters.
</para></listitem>
</varlistentry>
</variablelist>
</section>
<section>
<title>Example Cryptosystems</title>
<para>
In the <filename>example</filename> subdirectory there are various programs
that read curve parameters on standard input and perform computations
that would be required in a typical implementation of a pairing-based
cryptosystem. Sample schemes include:
</para>
<itemizedlist>
    <listitem>
    <para>
    Boneh-Franklin identity-based encryption
    </para>
    </listitem>
    <listitem>
    <para>
    Boneh-Lynn-Shacham short signatures
    </para>
    </listitem>
    <listitem>
    <para>
    Hess identity-based signatures
    </para>
    </listitem>
    <listitem>
    <para>
    Joux tripartite Diffie-Hellman
    </para>
    </listitem>
    <listitem>
    <para>
    Paterson identity-based signatures
    </para>
    </listitem>
    <listitem>
    <para>
    Yuan-Li identity-based authenticated key agreement
    </para>
    </listitem>
    <listitem>
    <para>
    Zhang-Kim identity-based blind/ring signatures
    </para>
    </listitem>
    <listitem>
    <para>
    Zhang-Safavi-Naini-Susilo signatures
    </para>
    </listitem>
</itemizedlist>
<para>
Of course, more work would be required to turn these programs
into real applications. Their main purpose is to showcase the PBC library
API.
</para>
</section>
<section>
<title>Benchmarks</title>
<para>
I use the programs in the <filename>benchmark</filename> subdirectory
to measure running times of pairings, and also RSA decryptions.
</para>
<para>
The <command>benchmark</command> program takes
pairing parameters on standard input and reports the average running
time of the pairing over 10 runs, while <command>timersa</command>
estimates the time required to perform one 1024-bit RSA decryption.
</para>
</section>
</chapter>