

浏览全部资源
扫码关注微信
南开大学数学系
Published:1981
移动端阅览
[1]沈世镒.具有边信息的信源编码定理[J].电子学报,1981(02):14-23.
Shen Shi-yi. The Source Coding Theorems with Side Information[J]. Acta Electronica Sinica, 1981, (2): 14-23.
本文第一部分给出了一般具有边信息信源序列S
n
为(R
n
D
n
)-压缩与(F
n
D
n
)-信息有界的定义
并且得到了具有边信息信源序列的编码基本定理:S
n
为(R
n
D
n
)-压缩的充要条件为
S
n
是(R
n
D
n
-信息有界。 由这个基本定理出发就可推出以下结论:1.一般信源序列S
n
关于率失真函数R
2
n
(D)的反编码定理;2.有边信息的平稳遍历信源关于R
2
(D)的正、反编码定理;3.对有边信息的平稳遍历信源关于R
2
(D)的反编码定理。
general source sequence with side information (ref. Fig-1).Definition 1: A source sequence with side information S(n)
is said to be (R(n)
D(n))-compound
if there exist δ<n)→0(δ(n)>
and a sequence Z0(n)Z(n)
such thati) P(n) (Z0(n))>l-δ(n);ii) For each z(n)∈Z0(n)
there exists a set B(n)(z(n))Y(n)
such that | B(n)(z(n)) | ≤2w where w = R(n)(1+δ(n))
and P{A(n)(z(n))/z(n)}>l-δ(n); where A(n)(z(n)) = {x(n)
P(n) [x(n)
B (n)(z(n))]≤D(n))}Definition 2: The S(n) is said to be (F(n)
D(n))-information bounded
if there exist e(n) →0
and conditional probability Q(n)(y(n)/x(n))
so that the joint distribution p(n) (x(n)
z(n)
wherei(x
y/z) = log[p(x
y/z)/p(x/z)p(y/z)Theorem 1: The necessary and sufficient condition for a source sequenc with side information S(n) to be (R(n)
D(n))-compound
is for the S(n) to be (R(n)
D(n))-information. bounded.From Theorem 1 and the law of large numbers
we may derive in an easier way the coding theorem for stationary or unstationary sources with memory.
0
Views
111
下载量
1
CSCD
Publicity Resources
Related Articles
Related Author
Related Institution
京公网安备11010802024621