prbs

2017-06-30

prbs

Library of pseudo-random binary sequence generators (LFSR-2 and LFSR-4) and related functions in Common Lisp.

Overview

This is a library of higher-order functions that will generate PRBS sequences of degree 3 through 786, 1024, 2048, and 4096. The closures produced by these functions will generate sequences as bitvectors, byte arrays, lists of bit-vectors, or lists of unsigned integers.

Other functions are included for detecting and tracking errors in PRBS data and finding the degree of unknown PRBS data.

Sequence Generation

The smallest PRBS sequence is PRBS-3, which is only 21 bits long.

(ql:quickload :prbs)
(use-package :prbs)

(funcall (bit-gen 3) 21)
=> #*010101011111110100001

Once a generator is created, subsequent calls to it will generate more data. Sequences repeat once they reach the end.

(defvar gen (num-gen 5))
=> GEN

(funcall gen 20)
=> (2 5 11 22 13 26 21 10 20 8 17 3 7 14 29 27 23 15 31 30)

(funcall gen 20)
=> (28 25 18 4 9 19 6 12 24 16 1 2 5 11 22 13 26 21 10 20)

Generators can be seeded with an integer value such as the current time in seconds. Otherwise they are seeded with the value 2, which makes the final value in the sequence always be 1. This is convenient for testing purposes but it makes early values in the sequence "zero heavy".

(defvar bytes (byte-gen 31))
=> BYTES

(funcall bytes 20)
=> #(0 0 0 4 0 0 0 16 0 0 0 64 0 0 1 0 0 0 4 0)

(setq bytes (byte-gen 31 :seed (get-universal-time)))
=> #<CLOSURE (LAMBDA (&OPTIONAL (PRBS::C 1)) :IN BYTE-GEN) {10030EA65B}>

(funcall bytes 20)
=> #(181 114 141 52 213 202 52 215 87 40 211 85 92 163 77 85 114 141 53 85)

Bit Error Detection

PRBS sequences are often used to characterize the error rate in communication links. If you generate packets of PRBS data and send them across an unreliable link, the received packets can be analyzed to determine the error rate of the comm link.

Assume that you have created packets of data for transmission:

(let ((gen (byte-gen 33 :seed (get-universal-time))))
     ...
     (transmit (funcall gen 40))
     ...
     (transmit (funcall gen 40))
     ...

In this case, 40-byte packets from PRBS-33 are being transmitted. Across the link, once any packet is received it can be used to "lock" on the PRBS sequence:

(ql:quickload :prbs)
(use-package :prbs.err)

(let ((tracker (prbs-lock (receive-packet-data ...) 33)))

The prbs-lock function will use the available data to locate where it is in the sequence and initialize a PRBS generator of its own that it uses to predict future data. Any arriving data that does not match the prediction is considered an error.

       (funcall tracker (receive-packet-data ...))
=> total-errors total-bits
       ...
       (funcall tracker (receive-packet-data ...))
=> total-errors total-bits
       ...

See sender.ros and recv.ros in the test folder for an example that does this with UDP datagrams.

Sequence Detection

Someday you might encounter some data that you suspect is from a PRBS, but you are not sure which one.

(ql:quickload :prbs)
(use-package '(prbs prbs.err prbs.util) :cl-user)

(prbs-detect (bytes->bits (loop repeat 75 collect (random 255))) :max 100)
=> NIL

(prbs-detect (take 600 (bit-gen 45 :seed (get-universal-time) :start 11)) :max 100)
=> (45)

As expected, a 75-byte sample of random data does not match any PRBS sequence. However, a 75-byte sample (600 bits) from PRBS-45 taken at an arbitrary non-aligned bit offset is enough to uniquely identify the sequence -- not bad considering the full PRBS-45 sequence is almost 200 terabytes long.

API Reference

API Reference

License

MIT

Author
Jason Lowdermilk <jlowdermilk@gmail.com>
License
MIT