line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
#!/usr/bin/perl |
2
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
package Net::OnlineCode; |
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
# play nicely as a CPAN module |
6
|
|
|
|
|
|
|
|
7
|
4
|
|
|
4
|
|
48863
|
use strict; |
|
4
|
|
|
|
|
10
|
|
|
4
|
|
|
|
|
185
|
|
8
|
4
|
|
|
4
|
|
24
|
use warnings; |
|
4
|
|
|
|
|
8
|
|
|
4
|
|
|
|
|
176
|
|
9
|
4
|
|
|
4
|
|
23
|
use vars qw(@ISA @EXPORT_OK @EXPORT %EXPORT_TAGS $VERSION); |
|
4
|
|
|
|
|
7
|
|
|
4
|
|
|
|
|
445
|
|
10
|
|
|
|
|
|
|
|
11
|
4
|
|
|
4
|
|
26
|
use constant DEBUG => 0; |
|
4
|
|
|
|
|
6
|
|
|
4
|
|
|
|
|
362
|
|
12
|
4
|
|
|
4
|
|
20
|
use constant ASSERT => 1; |
|
4
|
|
|
|
|
8
|
|
|
4
|
|
|
|
|
972
|
|
13
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
require Exporter; |
15
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
@ISA = qw(Exporter); |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
our @export_xor = qw (xor_strings safe_xor_strings fast_xor_strings); |
19
|
|
|
|
|
|
|
our @export_default = qw(); |
20
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
%EXPORT_TAGS = ( all => [ @export_default, @export_xor ], |
22
|
|
|
|
|
|
|
xor => [ @export_xor ], |
23
|
|
|
|
|
|
|
); |
24
|
|
|
|
|
|
|
@EXPORT_OK = ( @{ $EXPORT_TAGS{'all'} } ); |
25
|
|
|
|
|
|
|
@EXPORT = (); |
26
|
|
|
|
|
|
|
$VERSION = '0.03'; |
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
# Use XS for fast xors (TODO: make this optional) |
29
|
|
|
|
|
|
|
require XSLoader; |
30
|
|
|
|
|
|
|
XSLoader::load('Net::OnlineCode', $VERSION); |
31
|
|
|
|
|
|
|
|
32
|
|
|
|
|
|
|
# on to our stuff ... |
33
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
|
35
|
|
|
|
|
|
|
# Codec parameters |
36
|
|
|
|
|
|
|
# q is the number of message blocks that each auxiliary block will |
37
|
|
|
|
|
|
|
# link to |
38
|
|
|
|
|
|
|
# e epsilon, the degree of "suboptimality". Unlike Reed-Solomon or |
39
|
|
|
|
|
|
|
# Rabin's Information Dispersal Algorithm, Online Codes are not |
40
|
|
|
|
|
|
|
# optimal. This means that slightly more data needs to be generated |
41
|
|
|
|
|
|
|
# than either of these two codes. Also, whereas optimal codes |
42
|
|
|
|
|
|
|
# guarantee that a certain fraction of the "check" blocks/digits |
43
|
|
|
|
|
|
|
# suffice to reconstruct the original message, online codes only |
44
|
|
|
|
|
|
|
# guarantee that it can be reconstructed with a certain |
45
|
|
|
|
|
|
|
# probability |
46
|
|
|
|
|
|
|
# |
47
|
|
|
|
|
|
|
# Together with the number of blocks, n, these two variables define |
48
|
|
|
|
|
|
|
# the online code such that (1+qe)n check blocks are sufficient to |
49
|
|
|
|
|
|
|
# reconstruct the original message with a probability of 1 - (e/2) ** |
50
|
|
|
|
|
|
|
# (q+1). |
51
|
|
|
|
|
|
|
# |
52
|
|
|
|
|
|
|
|
53
|
4
|
|
|
4
|
|
25
|
use Carp; |
|
4
|
|
|
|
|
6
|
|
|
4
|
|
|
|
|
373
|
|
54
|
4
|
|
|
4
|
|
3040
|
use POSIX qw(ceil floor); |
|
4
|
|
|
|
|
34720
|
|
|
4
|
|
|
|
|
35
|
|
55
|
4
|
|
|
4
|
|
9068
|
use Digest::SHA qw(sha1 sha1_hex); |
|
4
|
|
|
|
|
16111
|
|
|
4
|
|
|
|
|
519
|
|
56
|
4
|
|
|
4
|
|
36
|
use Fcntl; |
|
4
|
|
|
|
|
9
|
|
|
4
|
|
|
|
|
13425
|
|
57
|
|
|
|
|
|
|
|
58
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
# Constructor for the base class |
60
|
|
|
|
|
|
|
# |
61
|
|
|
|
|
|
|
# This takes the three parameters that define the Online Code scheme, |
62
|
|
|
|
|
|
|
# corrects the value of epsilon if needed (see below) and then derives |
63
|
|
|
|
|
|
|
# the following: |
64
|
|
|
|
|
|
|
# |
65
|
|
|
|
|
|
|
# * max degree variable (F) |
66
|
|
|
|
|
|
|
# * number of auxiliary blocks (0.55 *qen) |
67
|
|
|
|
|
|
|
# * probability distribution p_1, p2, ... , p_F |
68
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
sub new { |
70
|
|
|
|
|
|
|
|
71
|
2
|
|
|
2
|
0
|
13
|
my $class = shift; |
72
|
|
|
|
|
|
|
# The default parameters used here for q and e (epsilon) are as |
73
|
|
|
|
|
|
|
# suggested in the paper "Rateless Codes and Big Downloads" by Petar |
74
|
|
|
|
|
|
|
# Maymounkov and David Maziere. Note that the value of e may be |
75
|
|
|
|
|
|
|
# overridden with a higher value if the lower value doesn't satisfy |
76
|
|
|
|
|
|
|
# max_degree(epsilon) > ceil(0.55 * q.e.mblocks) |
77
|
2
|
|
|
|
|
15
|
my %args = ( |
78
|
|
|
|
|
|
|
e => 0.01, |
79
|
|
|
|
|
|
|
q => 3, |
80
|
|
|
|
|
|
|
mblocks => undef, |
81
|
|
|
|
|
|
|
expand_aux => 0, |
82
|
|
|
|
|
|
|
e_warning => 0, |
83
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
# We don't use or store any RNG parameter that's been |
85
|
|
|
|
|
|
|
# passed into the constructor. |
86
|
|
|
|
|
|
|
@_ |
87
|
|
|
|
|
|
|
); |
88
|
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
|
90
|
2
|
|
|
|
|
8
|
my ($q,$e,$mblocks) = @args{qw(q e mblocks)}; |
91
|
|
|
|
|
|
|
|
92
|
2
|
50
|
|
|
|
7
|
unless (defined $args{mblocks}) { |
93
|
0
|
|
|
|
|
0
|
carp __PACKAGE__ . ": mblocks => (# message blocks) must be set\n"; |
94
|
0
|
|
|
|
|
0
|
return undef; |
95
|
|
|
|
|
|
|
} |
96
|
|
|
|
|
|
|
|
97
|
2
|
|
|
|
|
2
|
print "Net::OnlineCode mblocks = $mblocks\n" if DEBUG; |
98
|
|
|
|
|
|
|
|
99
|
2
|
|
|
|
|
3
|
my $P = undef; |
100
|
2
|
|
|
|
|
3
|
my $e_changed = 0; |
101
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
# how many auxiliary blocks would this scheme need? |
103
|
2
|
|
|
|
|
7
|
my $ablocks = _count_auxiliary($q,$e,$mblocks); |
104
|
|
|
|
|
|
|
|
105
|
|
|
|
|
|
|
# does epsilon value need updating? |
106
|
2
|
|
|
|
|
4
|
my $f = _max_degree($e); |
107
|
|
|
|
|
|
|
|
108
|
2
|
100
|
|
|
|
27
|
if ($f > $mblocks + $ablocks) { |
109
|
|
|
|
|
|
|
|
110
|
1
|
|
|
|
|
2
|
$e_changed = 1; |
111
|
|
|
|
|
|
|
|
112
|
1
|
50
|
|
|
|
3
|
if ($args{e_warning}) { |
113
|
0
|
|
|
|
|
0
|
print "E CHANGED!!\nWas: $e\n"; |
114
|
0
|
|
|
|
|
0
|
print "Gave F value of $f\n"; |
115
|
|
|
|
|
|
|
} |
116
|
|
|
|
|
|
|
|
117
|
|
|
|
|
|
|
# use a binary search to find a new epsilon such that |
118
|
|
|
|
|
|
|
# get_max_degree($epsilon) <= mblocks + ablocks (ie, n) |
119
|
1
|
|
|
|
|
2
|
my $epsilon = $e; |
120
|
|
|
|
|
|
|
|
121
|
|
|
|
|
|
|
local *eval_f = sub { |
122
|
11
|
|
|
11
|
|
6
|
my $t = shift; |
123
|
11
|
|
|
|
|
31
|
return _max_degree(1/(1 + exp(-$t))); |
124
|
1
|
|
|
|
|
5
|
}; |
125
|
|
|
|
|
|
|
|
126
|
1
|
|
|
|
|
4
|
my $l = -log(1/$e - 1); |
127
|
1
|
|
|
|
|
2
|
my $r = $l + 1; |
128
|
|
|
|
|
|
|
|
129
|
|
|
|
|
|
|
# expand right side of search until we get F <= n' |
130
|
1
|
|
|
|
|
2
|
while (eval_f($r) > $mblocks + $ablocks) { |
131
|
|
|
|
|
|
|
# $r = $l + ($r - $l) * 2; |
132
|
1
|
|
|
|
|
3
|
$r = 2 * $r - $l; |
133
|
|
|
|
|
|
|
} |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
# binary search between left and right to find a suitable lower |
136
|
|
|
|
|
|
|
# value of epsilon still satisfying F <= n' |
137
|
1
|
|
|
|
|
3
|
while ($r - $l > 0.01) { |
138
|
8
|
|
|
|
|
8
|
my $m = ($l + $r) / 2; |
139
|
8
|
100
|
|
|
|
10
|
if (eval_f($m) > $mblocks + $ablocks) { |
140
|
3
|
|
|
|
|
7
|
$l = $m; |
141
|
|
|
|
|
|
|
} else { |
142
|
5
|
|
|
|
|
8
|
$r = $m; |
143
|
|
|
|
|
|
|
} |
144
|
|
|
|
|
|
|
} |
145
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
# update e and ablocks |
147
|
1
|
|
|
|
|
3
|
$epsilon = 1/(1 + exp(-$r)); |
148
|
1
|
|
|
|
|
2
|
$f = eval_f($r); |
149
|
|
|
|
|
|
|
#$f=_max_degree($epsilon); |
150
|
1
|
50
|
|
|
|
3
|
carp __PACKAGE__ . ": increased epsilon value from $e to $epsilon\n" |
151
|
|
|
|
|
|
|
if $args{e_warning}; |
152
|
1
|
|
|
|
|
2
|
$e = $epsilon; |
153
|
1
|
|
|
|
|
2
|
$ablocks = _count_auxiliary($q,$e,$mblocks); |
154
|
|
|
|
|
|
|
|
155
|
1
|
50
|
|
|
|
5
|
if ($args{e_warning}) { |
156
|
|
|
|
|
|
|
|
157
|
0
|
|
|
|
|
0
|
print "Is now: $e\n"; |
158
|
0
|
|
|
|
|
0
|
print "New F: $f\n"; |
159
|
|
|
|
|
|
|
} |
160
|
|
|
|
|
|
|
|
161
|
|
|
|
|
|
|
} |
162
|
|
|
|
|
|
|
|
163
|
|
|
|
|
|
|
# calculate the probability distribution |
164
|
2
|
|
|
|
|
2
|
print "new: mblocks=$mblocks, ablocks=$ablocks, q=$q\n" if DEBUG; |
165
|
2
|
|
|
|
|
7
|
$P = _probability_distribution($mblocks + $ablocks,$e); |
166
|
|
|
|
|
|
|
|
167
|
2
|
50
|
|
|
|
9
|
die "Wrong number of elements in probability distribution (got " |
168
|
|
|
|
|
|
|
. scalar(@$P) . ", expecting $f)\n" |
169
|
|
|
|
|
|
|
unless @$P == $f; |
170
|
|
|
|
|
|
|
|
171
|
2
|
|
|
|
|
323
|
my $self = { q => $q, e => $e, f => $f, P => $P, |
172
|
|
|
|
|
|
|
mblocks => $mblocks, ablocks => $ablocks, |
173
|
|
|
|
|
|
|
coblocks => $mblocks + $ablocks, |
174
|
|
|
|
|
|
|
chblocks => 0, expand_aux=> $args{expand_aux}, |
175
|
|
|
|
|
|
|
e_changed => $e_changed, unique => {}, |
176
|
|
|
|
|
|
|
fisher_string => pack("L*", (0 .. $mblocks + $ablocks -1)), |
177
|
|
|
|
|
|
|
}; |
178
|
|
|
|
|
|
|
|
179
|
2
|
|
|
|
|
68
|
print "expand_aux => $self->{expand_aux}\n" if DEBUG; |
180
|
|
|
|
|
|
|
|
181
|
2
|
|
|
|
|
19
|
bless $self, $class; |
182
|
|
|
|
|
|
|
|
183
|
|
|
|
|
|
|
} |
184
|
|
|
|
|
|
|
|
185
|
|
|
|
|
|
|
# while it probably doesn't matter too much to the encoder whether the |
186
|
|
|
|
|
|
|
# supplied e value needed to be changed, if the receiver plugs the |
187
|
|
|
|
|
|
|
# received value of e into the constructor and it ends up changing, |
188
|
|
|
|
|
|
|
# there will be a problem with receiving the file. |
189
|
|
|
|
|
|
|
sub e_changed { |
190
|
0
|
|
|
0
|
0
|
0
|
return shift ->{e_changed}; |
191
|
|
|
|
|
|
|
} |
192
|
|
|
|
|
|
|
|
193
|
|
|
|
|
|
|
# convenience accessor functions |
194
|
|
|
|
|
|
|
sub get_mblocks { # count message blocks; passed into new |
195
|
2
|
|
|
2
|
0
|
11
|
return shift -> {mblocks}; |
196
|
|
|
|
|
|
|
} |
197
|
|
|
|
|
|
|
|
198
|
|
|
|
|
|
|
sub get_ablocks { # count auxiliary blocks; set in new |
199
|
2
|
|
|
2
|
0
|
12
|
return shift -> {ablocks}; |
200
|
|
|
|
|
|
|
} |
201
|
|
|
|
|
|
|
|
202
|
|
|
|
|
|
|
sub get_coblocks { # count composite blocks |
203
|
2
|
|
|
2
|
0
|
5
|
my $self = shift; |
204
|
2
|
|
|
|
|
12
|
return $self->{mblocks} + $self->{ablocks}; |
205
|
|
|
|
|
|
|
} |
206
|
|
|
|
|
|
|
|
207
|
|
|
|
|
|
|
# count checkblocks |
208
|
|
|
|
|
|
|
sub get_chblocks { |
209
|
0
|
|
|
0
|
0
|
0
|
return shift->{chblocks} |
210
|
|
|
|
|
|
|
} |
211
|
|
|
|
|
|
|
|
212
|
|
|
|
|
|
|
sub get_q { # q == reliability factor |
213
|
0
|
|
|
0
|
0
|
0
|
return shift -> {q}; |
214
|
|
|
|
|
|
|
} |
215
|
|
|
|
|
|
|
|
216
|
|
|
|
|
|
|
sub get_e { # e == suboptimality factor |
217
|
2
|
|
|
2
|
0
|
700
|
return shift -> {e}; |
218
|
|
|
|
|
|
|
} |
219
|
|
|
|
|
|
|
|
220
|
|
|
|
|
|
|
sub get_epsilon { # epsilon == e, as above |
221
|
0
|
|
|
0
|
0
|
0
|
return shift -> {e}; |
222
|
|
|
|
|
|
|
} |
223
|
|
|
|
|
|
|
|
224
|
|
|
|
|
|
|
sub get_f { # f == max (check block) degree |
225
|
2
|
|
|
2
|
0
|
12
|
return shift -> {f}; |
226
|
|
|
|
|
|
|
} |
227
|
|
|
|
|
|
|
|
228
|
|
|
|
|
|
|
sub get_P { # P == probability distribution |
229
|
0
|
|
|
0
|
0
|
0
|
return shift -> {P}; # (array ref) |
230
|
|
|
|
|
|
|
} |
231
|
|
|
|
|
|
|
|
232
|
|
|
|
|
|
|
|
233
|
|
|
|
|
|
|
# "Private" routines |
234
|
|
|
|
|
|
|
|
235
|
|
|
|
|
|
|
# calculate how many auxiliary blocks need to be generated for a given |
236
|
|
|
|
|
|
|
# code setup |
237
|
|
|
|
|
|
|
sub _count_auxiliary { |
238
|
3
|
|
|
3
|
|
3
|
my ($q, $e, $n) = @_; |
239
|
|
|
|
|
|
|
|
240
|
3
|
|
|
|
|
24
|
my $count = int(ceil(0.55 * $q * $e * $n)); |
241
|
3
|
|
|
|
|
4
|
my $delta = 0.55 * $e; |
242
|
|
|
|
|
|
|
|
243
|
3
|
|
|
|
|
4
|
warn "failure probability " . ($delta ** $q) . "\n" if DEBUG; |
244
|
|
|
|
|
|
|
#$count = int(ceil($q * $delta * $n)); |
245
|
|
|
|
|
|
|
|
246
|
|
|
|
|
|
|
# Is it better to change q or the number of aux blocks if q is too |
247
|
|
|
|
|
|
|
# big? It's certainly easier to keep the q value and increase the |
248
|
|
|
|
|
|
|
# number of aux blocks, as I'm doing here, and may even be the right |
249
|
|
|
|
|
|
|
# thing to do rather than ignoring the user's q value. |
250
|
3
|
50
|
|
|
|
6
|
if ($count < $q) { |
251
|
0
|
|
|
|
|
0
|
$count = $q; |
252
|
0
|
|
|
|
|
0
|
warn "updated _count_auxiliary output value to $q\n"; |
253
|
|
|
|
|
|
|
} |
254
|
3
|
|
|
|
|
5
|
return $count; |
255
|
|
|
|
|
|
|
} |
256
|
|
|
|
|
|
|
|
257
|
|
|
|
|
|
|
# The max degree specifies the maximum number of blocks to be XORed |
258
|
|
|
|
|
|
|
# together. This parameter is named F. |
259
|
|
|
|
|
|
|
sub _max_degree { |
260
|
|
|
|
|
|
|
|
261
|
15
|
|
|
15
|
|
11
|
my $epsilon = shift; |
262
|
|
|
|
|
|
|
|
263
|
15
|
|
|
|
|
35
|
my $quotient = (2 * log ($epsilon / 2)) / |
264
|
|
|
|
|
|
|
(log (1 - $epsilon / 2)); |
265
|
|
|
|
|
|
|
|
266
|
15
|
|
|
|
|
8
|
my $delta = 0.55 * $epsilon; |
267
|
|
|
|
|
|
|
#$quotient = (log ($epsilon) + log($delta)) / (log (1 - $epsilon)); |
268
|
|
|
|
|
|
|
|
269
|
15
|
|
|
|
|
36
|
return int(ceil($quotient)); |
270
|
|
|
|
|
|
|
} |
271
|
|
|
|
|
|
|
|
272
|
|
|
|
|
|
|
# Functions relating to probability distribution |
273
|
|
|
|
|
|
|
# |
274
|
|
|
|
|
|
|
# From the wikipedia page: |
275
|
|
|
|
|
|
|
# |
276
|
|
|
|
|
|
|
# http://en.wikipedia.org/wiki/Online_codes |
277
|
|
|
|
|
|
|
# |
278
|
|
|
|
|
|
|
# During the inner coding step the algorithm selects some number of |
279
|
|
|
|
|
|
|
# composite messages at random and XORs them together to form a check |
280
|
|
|
|
|
|
|
# block. In order for the algorithm to work correctly, both the number |
281
|
|
|
|
|
|
|
# of blocks to be XORed together and their distribution over composite |
282
|
|
|
|
|
|
|
# blocks must follow a particular probability distribution. |
283
|
|
|
|
|
|
|
# |
284
|
|
|
|
|
|
|
# Consult the references for the implementation details. |
285
|
|
|
|
|
|
|
# |
286
|
|
|
|
|
|
|
# The probability distribution is designed to map a random number in |
287
|
|
|
|
|
|
|
# the range [0,1) and return a degree i between 1 and F. The |
288
|
|
|
|
|
|
|
# probability distribution depends on a single input, n, which is the |
289
|
|
|
|
|
|
|
# number of blocks in the original message. The fixed values for q and |
290
|
|
|
|
|
|
|
# epsilon are also used. |
291
|
|
|
|
|
|
|
# |
292
|
|
|
|
|
|
|
# This code includes two changes from that described in the wikipedia |
293
|
|
|
|
|
|
|
# page. |
294
|
|
|
|
|
|
|
# |
295
|
|
|
|
|
|
|
# 1) Rather than returning an array of individual probabilities p_i, |
296
|
|
|
|
|
|
|
# the array includes the cumulative probabilities. For example, if |
297
|
|
|
|
|
|
|
# the p_i probabilities were: |
298
|
|
|
|
|
|
|
# (0.1, 0.2, 0.3, 0.2, 0.1, 0.1) |
299
|
|
|
|
|
|
|
# then the returned array would be: |
300
|
|
|
|
|
|
|
# (0.1, 0.3, 0.6, 0.8, 0.9, 1) (last element always has value 1) |
301
|
|
|
|
|
|
|
# This is done simply to make selecting a value based on the random |
302
|
|
|
|
|
|
|
# number more efficient, but the underlying probability distribution |
303
|
|
|
|
|
|
|
# is the same. |
304
|
|
|
|
|
|
|
# 2) Handling edge cases. These are: |
305
|
|
|
|
|
|
|
# a) the case where n = 1; and |
306
|
|
|
|
|
|
|
# b) the case where F > n |
307
|
|
|
|
|
|
|
# In both cases, the default value for epsilon cannot be used, so a |
308
|
|
|
|
|
|
|
# more suitable value is calculated. |
309
|
|
|
|
|
|
|
# |
310
|
|
|
|
|
|
|
# The return value is an array containing: |
311
|
|
|
|
|
|
|
# |
312
|
|
|
|
|
|
|
# * the max degree F |
313
|
|
|
|
|
|
|
# * a possibly updated value of epsilon |
314
|
|
|
|
|
|
|
# * the F values of the (cumulative) probability distribution |
315
|
|
|
|
|
|
|
|
316
|
|
|
|
|
|
|
sub _probability_distribution { |
317
|
|
|
|
|
|
|
|
318
|
2
|
|
|
2
|
|
2
|
my ($nblocks,$epsilon) = @_; # nblocks = number of *composite* blocks! |
319
|
|
|
|
|
|
|
|
320
|
|
|
|
|
|
|
# after code reorganisation, this shouldn't happen: |
321
|
2
|
50
|
|
|
|
5
|
if ($nblocks == 1) { |
322
|
0
|
|
|
|
|
0
|
croak "BUG: " . __PACKAGE__ ." - number of composite blocks = 1\n"; |
323
|
0
|
|
|
|
|
0
|
return (1, 0, 1); |
324
|
|
|
|
|
|
|
} |
325
|
|
|
|
|
|
|
|
326
|
2
|
|
|
|
|
3
|
print "generating probability distribution from nblocks $nblocks, e $epsilon\n" |
327
|
|
|
|
|
|
|
if DEBUG; |
328
|
|
|
|
|
|
|
|
329
|
2
|
|
|
|
|
2
|
my $f = _max_degree($epsilon); |
330
|
|
|
|
|
|
|
|
331
|
|
|
|
|
|
|
# after code reorganisation, this shouldn't happen: |
332
|
2
|
50
|
|
|
|
6
|
if ($f > $nblocks) { |
333
|
0
|
|
|
|
|
0
|
croak "BUG: " .__PACKAGE__ . " - epsilon still too small!\n"; |
334
|
|
|
|
|
|
|
} |
335
|
|
|
|
|
|
|
|
336
|
|
|
|
|
|
|
# probability distribution |
337
|
|
|
|
|
|
|
|
338
|
|
|
|
|
|
|
# Calculate the sum of the sequence: |
339
|
|
|
|
|
|
|
# |
340
|
|
|
|
|
|
|
# 1 + 1/F |
341
|
|
|
|
|
|
|
# p_1 = 1 - --------- |
342
|
|
|
|
|
|
|
# 1 + e |
343
|
|
|
|
|
|
|
# |
344
|
|
|
|
|
|
|
# |
345
|
|
|
|
|
|
|
# F . (1 - p_1) |
346
|
|
|
|
|
|
|
# p_i = --------------------- |
347
|
|
|
|
|
|
|
# (F - 1) . (i^2 - i) |
348
|
|
|
|
|
|
|
# |
349
|
|
|
|
|
|
|
# Since the i term is the only thing that changes for each p_i, I |
350
|
|
|
|
|
|
|
# optimise the calculation by keeping a fixed term involving only p |
351
|
|
|
|
|
|
|
# and f with a variable one involving i, then dividing as |
352
|
|
|
|
|
|
|
# appropriate. |
353
|
|
|
|
|
|
|
|
354
|
2
|
|
|
|
|
5
|
my $p1 = 1 - (1 + 1/$f)/(1 + $epsilon); |
355
|
2
|
|
|
|
|
4
|
my $pfterm = (1-$p1) * $f / ($f - 1); |
356
|
|
|
|
|
|
|
|
357
|
2
|
50
|
|
|
|
6
|
die "p1 is negative\n" if $p1 < 0; |
358
|
|
|
|
|
|
|
|
359
|
|
|
|
|
|
|
# hard-code simple cases where f = 1 or 2 |
360
|
2
|
50
|
|
|
|
9
|
if ($f == 1) { |
|
|
50
|
|
|
|
|
|
361
|
0
|
|
|
|
|
0
|
return [1]; |
362
|
|
|
|
|
|
|
} elsif ($f == 2) { |
363
|
0
|
|
|
|
|
0
|
return [$p1, 1]; |
364
|
|
|
|
|
|
|
} |
365
|
|
|
|
|
|
|
|
366
|
|
|
|
|
|
|
# calculate sum(p_i) for 2 <= i < F. |
367
|
|
|
|
|
|
|
# p_i=F is simply set to 1 to avoid rounding errors in the sum |
368
|
2
|
|
|
|
|
3
|
my $sum = $p1; |
369
|
2
|
|
|
|
|
3
|
my @P = ($sum); |
370
|
|
|
|
|
|
|
|
371
|
2
|
|
|
|
|
7
|
my $i = 2; |
372
|
2
|
|
|
|
|
6
|
while ($i < $f) { |
373
|
2515
|
|
|
|
|
1842
|
my $iterm = $i * ($i - 1); |
374
|
2515
|
|
|
|
|
1749
|
my $p_i = $pfterm / $iterm; |
375
|
|
|
|
|
|
|
|
376
|
2515
|
|
|
|
|
1605
|
$sum += $p_i; |
377
|
|
|
|
|
|
|
|
378
|
2515
|
50
|
|
|
|
3122
|
die "p_$i is negative\n" if $p_i < 0; |
379
|
|
|
|
|
|
|
|
380
|
2515
|
|
|
|
|
1967
|
push @P, $sum; |
381
|
2515
|
|
|
|
|
3282
|
$i++; |
382
|
|
|
|
|
|
|
} |
383
|
|
|
|
|
|
|
|
384
|
2
|
|
|
|
|
4
|
if (DEBUG) { |
385
|
|
|
|
|
|
|
# Make sure of the assumption that the sum of terms approaches 1. |
386
|
|
|
|
|
|
|
# If the "rounding error" below is not a very small number, we |
387
|
|
|
|
|
|
|
# know there is a problem with the assumption! |
388
|
|
|
|
|
|
|
my $p_last = $sum + $pfterm / ($f * $f - $f); |
389
|
|
|
|
|
|
|
my $absdiff = abs (1 - $p_last); |
390
|
|
|
|
|
|
|
warn "Absolute difference of 1,sum to p_F = $absdiff\n" if $absdiff >1e-8; |
391
|
|
|
|
|
|
|
} |
392
|
|
|
|
|
|
|
|
393
|
2
|
|
|
|
|
262
|
return [@P,1]; |
394
|
|
|
|
|
|
|
} |
395
|
|
|
|
|
|
|
|
396
|
|
|
|
|
|
|
|
397
|
|
|
|
|
|
|
# Fisher-Yates shuffle algorithm, based on recipe 4.17 from the Perl |
398
|
|
|
|
|
|
|
# Cookbook. Takes an input array it randomises the order (ie, |
399
|
|
|
|
|
|
|
# shuffles) and then truncates the array to "picks" elements. |
400
|
|
|
|
|
|
|
# |
401
|
|
|
|
|
|
|
# This is much more efficient than the usual approach of "keep picking |
402
|
|
|
|
|
|
|
# new elements until we get k distinct ones" particularly as k |
403
|
|
|
|
|
|
|
# approaches the size of the array. That algorithm could make |
404
|
|
|
|
|
|
|
# exponentially many calls to rand, whereas this just makes one call |
405
|
|
|
|
|
|
|
# per item to be picked. |
406
|
|
|
|
|
|
|
# |
407
|
|
|
|
|
|
|
# |
408
|
|
|
|
|
|
|
|
409
|
|
|
|
|
|
|
sub fisher_yates_shuffle { |
410
|
|
|
|
|
|
|
|
411
|
0
|
|
|
0
|
0
|
0
|
my ($rng, $array, $picks) = @_; |
412
|
|
|
|
|
|
|
|
413
|
0
|
|
|
|
|
0
|
if (ASSERT) { |
414
|
0
|
0
|
|
|
|
0
|
die "fisher_yates_shuffle: 1st arg not an RNG object\n" |
415
|
|
|
|
|
|
|
unless ref($rng); |
416
|
|
|
|
|
|
|
} |
417
|
|
|
|
|
|
|
|
418
|
|
|
|
|
|
|
# length in 32-bit words |
419
|
0
|
|
|
|
|
0
|
my $len = length($array) >> 2; |
420
|
|
|
|
|
|
|
|
421
|
|
|
|
|
|
|
# Change recipe to pick subset of list |
422
|
0
|
0
|
0
|
|
|
0
|
$picks=$len unless |
423
|
|
|
|
|
|
|
defined($picks) and $picks >= 0; |
424
|
|
|
|
|
|
|
|
425
|
|
|
|
|
|
|
# algorithm fills picks into the end of the array |
426
|
0
|
|
|
|
|
0
|
my $i=$len; |
427
|
0
|
|
|
|
|
0
|
while (--$i >= $len - $picks) { |
428
|
0
|
|
|
|
|
0
|
my $j=int($rng->rand($i + 1)); # range [0,$i] |
429
|
|
|
|
|
|
|
#next if $i==$j; # not worth checking, probably |
430
|
|
|
|
|
|
|
# @$array[$i,$j]=@$array[$j,$i] |
431
|
0
|
|
|
|
|
0
|
my $tmp1 = substr $array, $i << 2, 4; |
432
|
0
|
|
|
|
|
0
|
my $tmp2 = substr $array, $j << 2, 4; |
433
|
0
|
|
|
|
|
0
|
substr $array, $i << 2, 4, $tmp2; |
434
|
0
|
|
|
|
|
0
|
substr $array, $j << 2, 4, $tmp1; |
435
|
|
|
|
|
|
|
} |
436
|
|
|
|
|
|
|
|
437
|
0
|
|
|
|
|
0
|
return (unpack "L*", substr $array, ($len - $picks) << 2); |
438
|
|
|
|
|
|
|
} |
439
|
|
|
|
|
|
|
|
440
|
|
|
|
|
|
|
# |
441
|
|
|
|
|
|
|
# Routine to calculate the auxiliary block -> message block* mapping. |
442
|
|
|
|
|
|
|
# The passed rng object must already have been seeded, and both sender |
443
|
|
|
|
|
|
|
# and receiver should use the same seed. Returns [[..],[..],..] |
444
|
|
|
|
|
|
|
# representing which message blocks each of the auxiliary block is |
445
|
|
|
|
|
|
|
# composed of. |
446
|
|
|
|
|
|
|
# |
447
|
|
|
|
|
|
|
|
448
|
|
|
|
|
|
|
sub auxiliary_mapping { |
449
|
|
|
|
|
|
|
|
450
|
0
|
|
|
0
|
0
|
0
|
my $self = shift; |
451
|
0
|
|
|
|
|
0
|
my $rng = shift; |
452
|
|
|
|
|
|
|
|
453
|
0
|
0
|
|
|
|
0
|
croak "auxiliary_mapping: rng is not a reference\n" unless ref($rng); |
454
|
|
|
|
|
|
|
|
455
|
|
|
|
|
|
|
#print "auxiliary_mapping: entering RNG value: " . ($rng->as_hex). "\n"; |
456
|
|
|
|
|
|
|
|
457
|
|
|
|
|
|
|
# hash slices: powerful, but syntax is sometimes confusing |
458
|
0
|
|
|
|
|
0
|
my ($mblocks,$ablocks,$q) = @{$self}{"mblocks","ablocks","q"}; |
|
0
|
|
|
|
|
0
|
|
459
|
|
|
|
|
|
|
|
460
|
|
|
|
|
|
|
# make sure hash(ref) slice above actually did something sensible: |
461
|
|
|
|
|
|
|
# die "weird mblocks/ablocks" unless $nblocks + $aux_blocks >= 2; |
462
|
|
|
|
|
|
|
|
463
|
|
|
|
|
|
|
# I made a big mistake when reading the description for creating aux |
464
|
|
|
|
|
|
|
# blocks. What I implemented first (in the commented-out section |
465
|
|
|
|
|
|
|
# below) was to link each of the auxiliary blocks to q message |
466
|
|
|
|
|
|
|
# blocks. What I should have done was to link each *message block* |
467
|
|
|
|
|
|
|
# to q auxiliary blocks. As a result, it was taking much more than |
468
|
|
|
|
|
|
|
# the expected number of check blocks to decode the message. |
469
|
|
|
|
|
|
|
|
470
|
|
|
|
|
|
|
# as a result of the new algorithm, it makes sense to work out |
471
|
|
|
|
|
|
|
# reciprocal links between message blocks and auxiliary blocks |
472
|
|
|
|
|
|
|
# within the base class. Storing them here won't work out very well, |
473
|
|
|
|
|
|
|
# though: the encoder doesn't care about the message block to aux |
474
|
|
|
|
|
|
|
# block mapping, so it would be a waste of memory, but more |
475
|
|
|
|
|
|
|
# importantly, the decoder object stores all mappings in a private |
476
|
|
|
|
|
|
|
# GraphDecoder object (so duplicating the structure here would be a |
477
|
|
|
|
|
|
|
# waste). |
478
|
|
|
|
|
|
|
|
479
|
|
|
|
|
|
|
# I will make one change to the output, though: instead of just |
480
|
|
|
|
|
|
|
# returning the mappings for the 0.55qen auxiliary blocks, I will |
481
|
|
|
|
|
|
|
# return a list of message block *and* auxiliary block mappings. The |
482
|
|
|
|
|
|
|
# encoder and decoder will have to be changed: encoder immediately |
483
|
|
|
|
|
|
|
# splices the array to remove unwanted message block mappings, while |
484
|
|
|
|
|
|
|
# the decoder will be simplified by only having to pass the full |
485
|
|
|
|
|
|
|
# list to the graph decoder (which will have to be modified |
486
|
|
|
|
|
|
|
# appropriately). |
487
|
|
|
|
|
|
|
|
488
|
0
|
|
|
|
|
0
|
my $aux_mapping = []; |
489
|
|
|
|
|
|
|
|
490
|
0
|
|
|
|
|
0
|
my $ab_string = pack "L*", ($mblocks .. $mblocks + $ablocks -1); |
491
|
|
|
|
|
|
|
|
492
|
|
|
|
|
|
|
# list of empty hashes |
493
|
0
|
|
|
|
|
0
|
my @hashes; |
494
|
0
|
|
|
|
|
0
|
for (0 .. $mblocks + $ablocks -1) { $hashes[$_] = {}; } |
|
0
|
|
|
|
|
0
|
|
495
|
|
|
|
|
|
|
|
496
|
0
|
|
|
|
|
0
|
for my $msg (0 .. $mblocks - 1) { |
497
|
|
|
|
|
|
|
# list of all aux block indices |
498
|
|
|
|
|
|
|
|
499
|
0
|
|
|
|
|
0
|
foreach my $aux (fisher_yates_shuffle($rng, $ab_string, $q)) { |
500
|
0
|
|
|
|
|
0
|
$hashes[$aux]->{$msg}=undef; |
501
|
0
|
|
|
|
|
0
|
$hashes[$msg]->{$aux}=undef; |
502
|
|
|
|
|
|
|
} |
503
|
|
|
|
|
|
|
} |
504
|
|
|
|
|
|
|
|
505
|
|
|
|
|
|
|
# convert list of hashes into a list of lists |
506
|
0
|
|
|
|
|
0
|
for my $i (0 .. $mblocks + $ablocks -1) { |
507
|
0
|
|
|
|
|
0
|
print "map $i: " . (join " ", keys %{$hashes[$i]}) . "\n" if DEBUG; |
508
|
0
|
|
|
|
|
0
|
push @$aux_mapping, [ keys %{$hashes[$i]} ]; |
|
0
|
|
|
|
|
0
|
|
509
|
|
|
|
|
|
|
} |
510
|
|
|
|
|
|
|
|
511
|
|
|
|
|
|
|
# save and return aux_mapping |
512
|
0
|
|
|
|
|
0
|
$self->{aux_mapping} = $aux_mapping; |
513
|
|
|
|
|
|
|
} |
514
|
|
|
|
|
|
|
|
515
|
|
|
|
|
|
|
# Until I get the auto expand_aux working, this will have to do |
516
|
|
|
|
|
|
|
sub blklist_to_msglist { |
517
|
|
|
|
|
|
|
|
518
|
0
|
|
|
0
|
0
|
0
|
my ($self,@xor_list) = @_; |
519
|
|
|
|
|
|
|
|
520
|
0
|
|
|
|
|
0
|
my $mblocks = $self->{mblocks}; |
521
|
|
|
|
|
|
|
|
522
|
0
|
|
|
|
|
0
|
my %blocks; |
523
|
0
|
|
|
|
|
0
|
while (@xor_list) { |
524
|
0
|
|
|
|
|
0
|
my $entry = shift(@xor_list); |
525
|
0
|
0
|
|
|
|
0
|
if ($entry < $mblocks) { # is it a message block index? |
526
|
|
|
|
|
|
|
# toggle entry in the hash |
527
|
0
|
0
|
|
|
|
0
|
if (exists($blocks{$entry})) { |
528
|
0
|
|
|
|
|
0
|
delete $blocks{$entry}; |
529
|
|
|
|
|
|
|
} else { |
530
|
0
|
|
|
|
|
0
|
$blocks{$entry}= undef; |
531
|
|
|
|
|
|
|
} |
532
|
|
|
|
|
|
|
} else { |
533
|
|
|
|
|
|
|
# aux block : push all message blocks it's composed of |
534
|
0
|
|
|
|
|
0
|
my @expansion = @{$self->{aux_mapping}->[$entry]}; |
|
0
|
|
|
|
|
0
|
|
535
|
0
|
|
|
|
|
0
|
if (DEBUG) { |
536
|
|
|
|
|
|
|
print "expand_aux: expanding $entry to " . |
537
|
|
|
|
|
|
|
(join " ", @expansion) . "\n"; |
538
|
|
|
|
|
|
|
} |
539
|
0
|
|
|
|
|
0
|
push @xor_list, @expansion; |
540
|
|
|
|
|
|
|
} |
541
|
|
|
|
|
|
|
} |
542
|
0
|
|
|
|
|
0
|
return keys %blocks; |
543
|
|
|
|
|
|
|
} |
544
|
|
|
|
|
|
|
|
545
|
|
|
|
|
|
|
# Calculate the composition of a single check block based on the |
546
|
|
|
|
|
|
|
# supplied RNG. Returns a reference to a list of composite blocks |
547
|
|
|
|
|
|
|
# indices. |
548
|
|
|
|
|
|
|
|
549
|
|
|
|
|
|
|
sub checkblock_mapping { |
550
|
|
|
|
|
|
|
|
551
|
0
|
|
|
0
|
0
|
0
|
my $self = shift; |
552
|
0
|
|
|
|
|
0
|
my $rng = shift; |
553
|
|
|
|
|
|
|
|
554
|
0
|
0
|
|
|
|
0
|
croak "rng is not an object reference\n" unless ref($rng); |
555
|
|
|
|
|
|
|
|
556
|
0
|
|
|
|
|
0
|
my ($mblocks,$coblocks,$P) = @{$self}{"mblocks","coblocks","P"}; |
|
0
|
|
|
|
|
0
|
|
557
|
0
|
|
|
|
|
0
|
my @coblocks; |
558
|
|
|
|
|
|
|
|
559
|
|
|
|
|
|
|
# use weighted distribution to find how many blocks to link |
560
|
0
|
|
|
|
|
0
|
my $i = 0; |
561
|
0
|
|
|
|
|
0
|
my $r = $rng->rand; |
562
|
0
|
|
|
|
|
0
|
++$i while($r > $P->[$i]); # terminates since r < P[last] |
563
|
0
|
|
|
|
|
0
|
++$i; |
564
|
|
|
|
|
|
|
|
565
|
|
|
|
|
|
|
# select i composite blocks uniformly |
566
|
0
|
|
|
|
|
0
|
@coblocks = fisher_yates_shuffle($rng, $self->{fisher_string} , $i); |
567
|
|
|
|
|
|
|
|
568
|
0
|
|
|
|
|
0
|
if (ASSERT) { |
569
|
0
|
0
|
|
|
|
0
|
die "fisher_yates_shuffle: created empty check block\n!" unless @coblocks; |
570
|
|
|
|
|
|
|
} |
571
|
|
|
|
|
|
|
|
572
|
0
|
|
|
|
|
0
|
print "CHECKblock mapping: " . (join " ", @coblocks) . "\n" if DEBUG; |
573
|
|
|
|
|
|
|
|
574
|
0
|
|
|
|
|
0
|
return \@coblocks; |
575
|
|
|
|
|
|
|
|
576
|
|
|
|
|
|
|
} |
577
|
|
|
|
|
|
|
|
578
|
|
|
|
|
|
|
# non-method sub for xoring a source string (passed by reference) with |
579
|
|
|
|
|
|
|
# one or more target strings. I may reimplement this using XS later to |
580
|
|
|
|
|
|
|
# make it more efficient, but will keep a pure-perl version with this |
581
|
|
|
|
|
|
|
# name. |
582
|
|
|
|
|
|
|
sub safe_xor_strings { |
583
|
|
|
|
|
|
|
|
584
|
4
|
|
|
4
|
0
|
25
|
my $source = shift; |
585
|
|
|
|
|
|
|
|
586
|
|
|
|
|
|
|
# die if user forgot to pass in a reference (to allow updating) or |
587
|
|
|
|
|
|
|
# called $self->safe_xor_strings by mistake |
588
|
4
|
50
|
|
|
|
18
|
croak "xor_strings: arg 1 should be a reference to a SCALAR!\n" |
589
|
|
|
|
|
|
|
unless ref($source) eq "SCALAR"; |
590
|
|
|
|
|
|
|
|
591
|
4
|
|
|
|
|
8
|
my $len = length ($$source); |
592
|
|
|
|
|
|
|
|
593
|
4
|
50
|
|
|
|
13
|
croak "xor_strings: source string can't have zero length!\n" |
594
|
|
|
|
|
|
|
unless $len; |
595
|
|
|
|
|
|
|
|
596
|
4
|
|
|
|
|
10
|
foreach my $target (@_) { |
597
|
4
|
50
|
|
|
|
15
|
croak "xor_strings: targets not all same size as source\n" |
598
|
|
|
|
|
|
|
unless length($target) == $len; |
599
|
4
|
|
|
|
|
11
|
map { substr ($$source, $_, 1) ^= substr ($target, $_, 1) } |
|
28
|
|
|
|
|
95
|
|
600
|
|
|
|
|
|
|
(0 .. $len-1); |
601
|
|
|
|
|
|
|
} |
602
|
|
|
|
|
|
|
|
603
|
4
|
|
|
|
|
24
|
return $$source; |
604
|
|
|
|
|
|
|
} |
605
|
|
|
|
|
|
|
|
606
|
|
|
|
|
|
|
# Later, xor_strings could be replaced with an C version with reduced |
607
|
|
|
|
|
|
|
# error checking, so make a backward-compatible version and an |
608
|
|
|
|
|
|
|
# explicit fast/unsafe version. |
609
|
0
|
|
|
0
|
0
|
|
sub xor_strings { safe_xor_strings(@_) } |
610
|
|
|
|
|
|
|
#sub fast_xor_strings { safe_xor_strings(@_) } # implemented in OnlineCode.xs. |
611
|
|
|
|
|
|
|
|
612
|
|
|
|
|
|
|
|
613
|
|
|
|
|
|
|
1; |
614
|
|
|
|
|
|
|
|
615
|
|
|
|
|
|
|
__END__ |