|
|
Thresholds for terative Equalization of Partial Response Channels
Using Density Evolution
|
ISIT 2001
Washington D.C.
June, 2001
|
|
J.Li,
K.R.Narayanan, C.N.Georghades, E.Kurtas
|
|
|
»
full
[ps]
[pdf] short
[ps]
[pdf]
Abstract --
This paper focuses on the computing of the thresholds for iterative equalization and decoding of partial response
channel, such as what are of interest in high density magnetic recording.
Extensive simulations on the performance of turbo codes, low density parity check (LDPC) codes and
single-parity check turbo product codes (TPC/SPC) over partial response channels reveal very encouraging
bit error rate (BER) performance, but cast little insight into the fundamental limitation within the existing suboptimal
decoding. This work uses the method of density evolution with Gaussian approximation to investigate the
achievable capacity by taking into consideration of both the code's structural properties and the iterative
decoding algorithm. A unified framework involving different outer codes, including (punctured) convolutional
codes (also known as serial turbo), LDPC codes and TPC/SPC codes, is presented. Certain interesting
issues in the decoding optimization are investigated. The tightness demonstrated by
the bound attributes density evolution to be a very useful tool to be extended for
capacity analysis of a broad range of iterative processes.
Keywords --
performance bounds,
turbo codes, Low density parity check (LDPC) codes, turbo product codes (TPC),
density evolution, inter-symbol interference (ISI) channels,
partial-response (PR) channels, message-passing decoding,
turbo equalization,
data storage systems
|
|
|