Compression scheme for interval data
View Patent ↗A method of data compression for use in a data communication's system for an electrical utility or the like to affect the most efficient transfer of data. The data is first formed into a forming a block (B). A parameter (M) is selected from among a group of parameters. Each data value in the block is divided by the selected parameter to produce, for each data value, a quotient (q=└n/M┘) and a remainder (r=n−qM). The quotient of each data value is encoded using unary coding in which the value for the quotient is encoded by transmitting an appropriate number of zeroes followed by a one. The remainder of each data value is encoded using binary encoding, and the resultant encoded block of data is then transmitted. The data values comprising the block of data are positive integer values and the encoding employs a Golomb-Rice code defined by a positive integer parameter.
1. A method of data compression for affecting the most efficient transfer of data comprising:
collecting data to be transmitted from one site to another site, and providing the data to a processor, the processor forming the data into a block of data and encoding the block of data for transmission; and,
the processor, before the block of data is encoded, pre-conditioning the data by finding a minimum value of the data comprising the block and subtracting that value from all other data values in the block while leaving the minimum value of data in the block unchanged, pre-conditioning the data serving to increase data coding efficiency.
2. The method of claim 1 in which the processor encodes the data using a Golomb-Rice code.
3. The method of claim 1 in which the processor transmits the resultant encoded block of data from the one site to the other site, including separately transmitting an index value for the block of data.
4. The method of claim 3 in which the separately transmitted index is encoded using log 2 N bits for a block of data containing N data values.
5. The method of claim 1 further including the processor selecting a parameter from a group of parameters known to the receiver, dividing each pre-conditioned data value in the block by the selected parameter to produce, for each data value, a quotient and a remainder, the pre-conditioned data having positive integer values n, encoding of the data being accomplished using a code defined by a positive integer parameter M with each value n being divided by M to produce a quotient q=└n/M┘, and a remainder r=n−qM; where the parameter is chosen to maximize the compression of the encoded data.
6. The method of claim 5 in which the processor encodes the quotient of each pre-conditioned data value using unary coding and the remainder of each value thereof using binary encoding.
7. The method of claim 5 in which the divisor for dividing each pre-conditioned data value is determined by the processor making a single pass over the pre-conditioned data comprising the block of data using a defined algorithm.
8. The method of claim 7 in which the algorithm is defined by computing a histogram function h i of the B data values to be encoded, where x b,0 ,x b,1 , . . . x b,7 represent the 8 most significant bits, with x 0 being the least significant bit, of the b th value to be encoded, and h i is defined as
h
i
=
∑
b
=
1
B
x
b
,
i
,
and is computed from the single pass over the data.
9. The method of claim 8 in which the parameter M is set to be 2 R−8+L , where R is the number of bits of the uncompressed data and L is the value that minimizes the function:
B
+
∑
i
=
L
7
2
i
-
L
h
i
,
which can be determined efficiently by initially setting L to zero, evaluating the quantity
H
L
=
h
L
+
2
-
L
-
1
∑
i
=
L
+
1
7
2
i
h
i
,
incrementing L if H L <B, and repeating the calculation of H L for increasing values of L, until H L ≧B, and setting L to be the largest value for which H L <B.
10. The method of claim 5 in which the parameter M is estimated from a block of B data values in a single pass over the pre-conditioned data by finding x max , the largest value in the block of data, and setting M to 2 P−3 , where P is the location of the most significant bit of x max containing a logical “1”.
11. A method of data compression for affecting the most efficient transfer of data comprising:
collecting data to be transmitted from one site to another site, and providing the data to a processor, the processor forming the data into a block of data and encoding the block of data for transmission; and,
the processor, before the block of data is encoded, locating the maximum value x max in the block of data;
finding the location of the most significant bit of x max containing a logical “1” and,
truncating a fixed number of bits from the most significant bits of all of the values in the block of data such that the most significant bit of the maximum value is a logical “1”.
12. The method of claim 11 in which the processor transmits the truncated binary values, together with a value indicating the number of leading zeros truncated from each of the encoded values.