MT Convolutional Decoder (Viterbi Soft Decision, Generator Matrix)

input integers

Bit metrics for quantized demodulated BPSK symbol data. These demodulated BPSK symbols are quantized to n_soft bits of precision, such that an integer value of 0 represents a strong logic 0 and a value of 2ⁿ_soft represents a strong logic 1.

Note To generate integer-valued quantized BPSK demodulated data from the Modulation nodes, decimate and quantize the oversampled data at the output of the PSK demodulator/matched filter.

For higher-order M-PSK/QAM schemes, convert the I/Q symbol values into corresponding quantized bit metrics using a suitable algorithm/heuristic.

Default value: 1

generator matrix

The generator connection polynomial matrix that sets the convolutional feedforward node connections in octal format.

The convolutional node is modeled as a linear feedforward shift register arrangement consisting of k rows with K-1 shift registers per row, where k denotes the data word length and K denotes the constraint length. If a_ij {0 ≤ i ≤ n-1, 0 ≤ j ≤ k-1} denotes a particular element in the generator matrix, the row index i corresponds to the convolutional node output y_i that is affected by this element, while the column index j corresponds to the jth row in the k row shift register arrangement. Thus a_ij specifies how the K bits in the jth row of the feedforward shift register matrix affects the ith output of the convolutional node.

For a rate of 2/3, the generator matrix is specified as follows:

The matrix on the right represents the elements in octal format. Zeros are padded at the end of the corresponding code generator sequences such that their total length is a multiple of three digits. The following diagram depicts the rate 2/3 convolutional node corresponding to the previously mentioned generator matrix, with a constraint length equal to 4. In the following diagram, D represents a shift register or memory element.

1378

Here, y_i^j, 0 ≤ j ≤ n-1 denotes the jth output of this node, in the ith instance.

Default value:

constraint length

The maximum number of encoded bits that can be affected by a single input bit. This value represents (1 + maximal memory order), where maximal memory order refers to the length of the longest shift register chain in the convolutional encoder.

Default value: 3

initial state

The initial parent state for the decode operation. When reset? is set to TRUE, the trellis structure is set to this state, thereby initializing the Viterbi decoding operation. On the first call to this node, and thereafter when reset? is set to FALSE, the survivor state from the previous iteration is used to continue performing Viterbi decoding and this parameter is ignored.

Default value: 0

error in

Error conditions that occur before this node runs.

The node responds to this input according to standard error behavior.

Standard Error Behavior

Default value: No error

traceback depth

The number of trellis stages used in the Viterbi decoding process.

Default value: 15

soft decision bits

The number of bits used for quantizing the BPSK modulated symbols to integers.

Default value: 1

reset?

A Boolean that determines whether the internal state of the decoder is cleared.

TRUE

Clears any buffered bits from previous iterations. Also initializes the Viterbi algorithm to start decoding from initial state. The relationship of the length of output bit stream, L_out, to the length of input bit stream, L_in, is described by the following equation:

L_out = k × [floor(L_in/n) - D]

where k is the input data word length, n is the output data word length, and D is the decoder traceback depth in symbols.

FALSE

Continues decoding from the previous iteration. The length of output bit stream is given by the following equation:

L_out = k × floor(L_in/n).

Tip If the length of the encoded bit stream is L_in, and reset? is set to TRUE, the Viterbi decoding algorithm (for a rate k/n code) returns a total of L_out = k×[floor(L_in/n)-D] decoded bits in a single iteration, implying that a total of k×D message bits are buffered inside the node. To recover the entire message of length k×floor[L_in/n] in a single call to MT Convolutional Decoder with reset? set to TRUE, choose one of the following options:

Append max(D,K-1)×k zeros to the end of the message prior to performing convolutional encoding. Here k represents the message word length.
Append D×n zeros to the end of the encoded bit stream prior to calling MT Convolutional Decoder. Here, D is the traceback depth employed in the convolutional decoder, and n is the code word length.

Default value: TRUE

output bit stream

Bit sequence decoded by this node.

final state

The value for the k(K-1) shift registers as the right-aligned (least significant) k(K-1) bits when this node completes execution, where K is the constraint length and k is the data word length in bits.

error out

Error information.

The node produces this output according to standard error behavior.