IS-95A Fwd Ch Detector (CDMA Reference Blockset)

Perform despreading, demodulation, and rake combining

Library

IS-95A Mobile Station Receiver

Description

This is a hierarchical block consisting of four rake fingers, coherent rake demodulator, descrambler, and short and the long PN code generators. This block performs the front-end forward channel receiver functions to detect the received signal. For each finger, this block downsamples the received signal, despreads it with the Walsh sequence, correlates it with the short PN code, and estimates the pilot channel multipath strength at the phase assigned to the finger. Using the channel estimates for each finger, this block demodulates the received signal, extracts the power bits, and combines the signals from all of the rake fingers to represent the soft decision detected symbols whose value is scrambled with the decimated long code. Finally, this block buffers and outputs a frame of detected soft-decision symbols after descrambling them with the decimated long code.

Inputs

Walsh Seq: Real vector of bipolar data representing the Walsh code sequence. If the Walsh order parameter is W, then the Walsh Seq input vector size is 2^W.
Rx Signal: Complex vector representing the oversampled input signal. The vector size equals the Walsh Seq input vector size times the Oversampling rate parameter times the Input frame size parameter.
Path Delay/Enable: Integer vector of size eight containing a pair of elements for each rake finger. The first element in each pair represents the initial PN phase offset that is applied to each rake finger. The second element in each pair is an enable signal for the rake finger. A value of 0 disables the finger and a value of 1 enables the finger.
Short PN Mask: Integer vector of size two representing the masks for the in-phase and quadrature components of the short PN code. The masks instantaneously change the phase value of the short PN code generator output.

Outputs

Real vector of size 384 representing an output frame of detected symbols.

Dialog Box

Parameters

Rate set: The rate set, either Rate Set I or Rate Set II.
Walsh order: Integer scalar that specifies the order of the Walsh code used to encode each data symbol.
Input frame size (in number of symbols): Integer scalar that specifies the number of symbols to be demodulated in each execution of the block.
Oversampling rate: Integer scalar that specifies the number of samples per chip. The default value, 8, corresponds to eight samples in a chip interval.
Tracking buffer size (in number of symbols): Integer scalar that specifies the buffer length (in number of symbols) required to correct for differences in path delays of the received multipath rays. The length of this buffer sets a limit on the maximum phase difference between the received rays.
Sample time: Real scalar that specifies the block sample time.

Algorithm

The signal received by the rake receiver module of the mobile can be modeled by

where:

k is the time index
t is between kT and (k+1)T
T is the period of a chip (1/1228800 second)
L is the number of propagation delay paths
J is the number of users on the same CDMA channel
(k,l) is the equivalent complex gain of the l-th delay path of the propagation channel at sample k
(l) is the delay of path l
{b(k,j)} is the jth user's Walsh-spread sequence
p(k) is the short (complex) PN sequence that identifies the base station

Furthermore, z(t) models the interference generated by other CDMA base stations, while (t) models the thermal noise, as well as the existing narrow-band interference.

b(k,j), in turn, can be expanded as

= s

where:

m = k (mod 64)
n = (k-m)/64
{w(m,j)} is the Walsh sequence
{s(n,j)} is the sequence of bipolar encoded data bits sent by the base station to the jth user/channel

Because different Walsh sequences are orthogonal and the short PN sequence has zero cross-correlation over its period, the optimum way to detect the transmitted data bits {s(n,j)} sent over the propagation-delay path l, is to multiply the received signal by p*(k)w(m,j)u(t-kT-(l)) (where p* is the conjugate of p) and integrate over the Walsh sequence period, that is, 64 chips. This, in fact, is the principle behind the construction of the rake receiver. After synchronizing each finger to a different propagation-delay path, the output of lth finger for the jth user (or channel), is then given by

where I(n,d), Z(n), and E(n) are respectively the residual terms due to transmission through other delay paths, interference from other base stations, and noise. If one could integrate over the whole period of the PN sequence, then all the residual terms would be close to zero (assuming that the transmitted bits spread by the Walsh sequence have white noise characteristics). However, due to the limited integration length, these terms have small but nevertheless nonzero values.

Although none of the three above-mentioned interference terms is statistically independent of the first term, for all practical purposes one can assume that they behave like white noise. Moreover, because the rate of variation of propagation channel characteristics, that is, (k,l), is in the order of the Doppler frequency and is much smaller than the symbol rate (19.2 kbps), one can assume in practice that (k,l) is constant during the integration period equal to 64 chips or one symbol length. Hence, one can approximate the above equation with

where e(n,l,j) is assumed to be a white noise sequence and (n,l,j) reflects the variations of the propagation channel characteristics, and therefore, is slowly varying.

The objective of the demodulator, therefore, is to calculate the optimum estimate

(or soft-detection) of s(n,j) for each finger, and to combine these estimates so as to maximize the resulting signal-to-interference ratio.

To combine the demodulated data from different rake fingers, simply add the soft-detected symbols together. In other words, the symbol representing soft detection of s(n,j) is given by

where L is the number of rake fingers.

This is based on the assumption that the distribution functions of the rake finger outputs are independent, which means that the joint log-likelihood function of the outputs is simply the sum of the individual log-likelihood functions. If the time-alignments of different fingers are sufficiently far apart, then this is a valid assumption.

Specification References

IS-95A 7.1.3.1.8, 7.1.3.1.9
J-STD-008 3.1.3.1.9, 3.1.3.1.10

IS-95A Fwd Ch Descrambler IS-95A Fwd Ch Interleaver/Deinterleaver