How big is the problem of recording all telephone conversations

[David Farber <farber () central cis upenn edu>]

From: Gerald Maguire <maguire () it kth se>
Date: Sun, 5 Mar 1995 01:04:30 +0100
To: farber () central cis upenn edu
Subject: recording


I just posted the following - since I though about our discussion and
did the calculations again yesterday:


--------------------------------------------------------------------------


Is recording all calls feasible?


To put it in perspective:


for 10^8 lines at 100 minutes per day using 4T byte tapes (such as
might be feasible from:
        CREO products Inc.
        110-3700 Gilmore Way
        Burnaby, B.C. V5G 4M1
        Canada
 [note they sold 1Tbyte tapes several years ago])


with 100 centers recording calls [~the number of main interexchange
centers in the US]


digitizing at 1 kbyte/sec [assuming reasonable LPC encoding, etc.]
leads to 1.54 tapes per day per center or 560.64 tapes per year per
center.


But what if you went to higher data rates [for example 64Kbits/sec of
an ISDN B channel, and 24 hours per day of traffic. It is still
less that 177 tapes/day/center or less than 65,000 tapes per center
per year.


Since the true capacity needs lies somewhere between these two
extremes (both due to data rates and due to call minutes/day) - it
seems feasible to record all telephone calls in a country the size of
the U.S. This is not to say that it is done, just that it is feasible
[especially if each exchange must have facilites to provide redirects
of a copy of any call - why not redirect them all].


Is it feasible to handle all these tapes?
        Many physical experiments generate 2,000 or more reels of tape per experiment


        Lots of other systems generate comparable data volumes (for
                example 50Mbits/s satellite sensors ==> ~4*10^3
                Gbytes/day  -  comparable to about 1 minute of calling
                per phone for 10^8 phones)


Chip
----------------------------------------------------------------------


[see comments at the end!! djf]


Here are the numbers etc. in DIF format to load into your favorite
spreadsheet program:


TABLE
0,1
""
VECTORS
0,11
""
TUPLES
0,23
""
DATA
0,0
""
-1,0
BOT
1,0
"US"
1,0
"one per every 2 persons"
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
-1,0
BOT
1,0
"Number of lines"
1,0
"Number of minutes of calls"
1,0
"Bytes/second"
1,0
"Mbyes/day/line"
1,0
"Mbytes/monthly/line"
1,0
"MBytes/year- per line"
1,0
"Total Gbytes/day"
1,0
"Number of centers (distrinuted ecording)"
1,0
"GBytes/day/center"
1,0
"4T tapes/day/Center"
1,0
"4T Tapes/year"
-1,0
BOT
0,100000000
V
0,1
V
0,1024
V
0,0.06144
V
0,1.8432
V
0,22.4256
V
0,6144
V
0,100
V
0,61.44
V
0,0.01536
V
0,5.6064
V
-1,0
BOT
0,100000000
V
0,10
V
0,1024
V
0,0.6144
V
0,18.432
V
0,224.256
V
0,61440
V
0,100
V
0,614.4
V
0,0.1536
V
0,56.064
V
-1,0
BOT
0,100000000
V
0,100
V
0,1024
V
0,6.144
V
0,184.32
V
0,2242.56
V
0,614400
V
0,100
V
0,6144
V
0,1.536
V
0,560.64
V
-1,0
BOT
0,100000000
V
0,1440
V
0,1024
V
0,88.4736
V
0,2654.208
V
0,32292.864
V
0,8847360
V
0,100
V
0,88473.6
V
0,22.1184
V
0,8073.216
V
-1,0
BOT
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
0,0
V
-1,0
BOT
0,100000000
V
0,1
V
0,8192
V
0,0.49152
V
0,14.7456
V
0,179.4048
V
0,49152
V
0,100
V
0,491.52
V
0,0.12288
V
0,44.8512
V
-1,0
BOT
0,100000000
V
0,10
V
0,8192
V
0,4.9152
V
0,147.456
V
0,1794.048
V
0,491520
V
0,100
V
0,4915.2
V
0,1.2288
V
0,448.512
V
-1,0
BOT
0,100000000
V
0,100
V
0,8192
V
0,49.152
V
0,1474.56
V
0,17940.48
V
0,4915200
V
0,100
V
0,49152
V
0,12.288
V
0,4485.12
V
-1,0
BOT
0,100000000
V
0,1440
V
0,8192
V
0,707.7888
V
0,21233.664
V
0,258342.912
V
0,70778880
V
0,100
V
0,707788.8
V
0,176.9472
V
0,64585.728
V
-1,0
BOT
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
-1,0
BOT
1,0
"Sweden"
1,0
"one per person"
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
-1,0
BOT
1,0
"Number of lines"
1,0
"Number of minutes of calls"
1,0
"Bytes/second"
1,0
"Mbyes/day/line"
1,0
"Mbytes/monthly/line"
1,0
"MBytes/year- per line"
1,0
"Total Gbytes/day"
1,0
"Number of centers (distrinuted ecording)"
1,0
"GBytes/day/center"
1,0
"4T tapes/day/Center"
0,0
V
-1,0
BOT
0,10000000
V
0,1
V
0,1024
V
0,0.06144
V
0,1.8432
V
0,22.4256
V
0,614.4
V
0,100
V
0,6.144
V
0,0.001536
V
0,0.56064
V
-1,0
BOT
0,10000000
V
0,10
V
0,1024
V
0,0.6144
V
0,18.432
V
0,224.256
V
0,6144
V
0,100
V
0,61.44
V
0,0.01536
V
0,5.6064
V
-1,0
BOT
0,10000000
V
0,100
V
0,1024
V
0,6.144
V
0,184.32
V
0,2242.56
V
0,61440
V
0,100
V
0,614.4
V
0,0.1536
V
0,56.064
V
-1,0
BOT
0,10000000
V
0,1440
V
0,1024
V
0,88.4736
V
0,2654.208
V
0,32292.864
V
0,884736
V
0,100
V
0,8847.36
V
0,2.21184
V
0,807.3216
V
-1,0
BOT
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
1,0
""
-1,0
BOT
0,10000000
V
0,1
V
0,8192
V
0,0.49152
V
0,14.7456
V
0,179.4048
V
0,4915.2
V
0,100
V
0,49.152
V
0,0.012288
V
0,4.48512
V
-1,0
BOT
0,10000000
V
0,10
V
0,8192
V
0,4.9152
V
0,147.456
V
0,1794.048
V
0,49152
V
0,100
V
0,491.52
V
0,0.12288
V
0,44.8512
V
-1,0
BOT
0,10000000
V
0,100
V
0,8192
V
0,49.152
V
0,1474.56
V
0,17940.48
V
0,491520
V
0,100
V
0,4915.2
V
0,1.2288
V
0,448.512
V
-1,0
BOT
0,10000000
V
0,1440
V
0,8192
V
0,707.7888
V
0,21233.664
V
0,258342.912
V
0,7077888
V
0,100
V
0,70778.88
V
0,17.69472
V
0,6458.5728
V
-1,0
EOD


----------------------------------------------------------------------


The suprise is that even going to 24 hours/phone/day and a complete B
channel (with
no coding) - the data volume is not infeasible. It would be modestly
expensive (about $260M/year for tapes [the 1Tbyte tapes were roughly
$4k each, but given media prices it should now be about 4T for the
same price]. The drives used to cose $250K each, the cost of drives is
about 2x the cost of the tapes, thus leading to less tha $1B for 10^8
B channels 24 hours per day! [In some circles this would be viewed as
a to small to merit an line of its own for appropriation purposes.]


Enjoy,
Chip