Data transmission

Data system —► Coder —Modulator —► Transmitter

Receinng station or relay orbiter

Data system Decoder Demodulator Receiver


Figure 10.5. Diagram of data transmission.

frequency during entry and descent, and a small link margin, especially in DTE links, often makes a subcarrier with residual carrier modulation preferable; it is much easier, for example, to perform Doppler tracking on the unsuppressed carrier signal than on the modulated subcarrier with suppressed carrier.

The bit error rate (BER) is the main parameter characterizing the performance of a communication link. The ratio of the energy per bit of raw data to the noise spectral density Eb/N0 at a given BER characterizes the efficiency of coding and modulation. Essentially what coding does (as with error detection and correction codes for memories exposed to radiation) is to add redundancy to the data by adding symbols. Although this increases the total number of bits to be transmitted, the error-correction ability more than compensates, such that the probability of uncorrectable errors is reduced overall (or the transmission rate for a given power can be increased; sometimes the effect is expressed as a 'coding gain' - the boost in power that would give the same improvement in data rate for a given BER). The penalty is in the additional hardware and/or software required for both the transmitter and receiver - dedicated hardware has historically been used, although software-based coders and decoders are becoming more common as the algorithms become more complex.

Uncoded PCM (pulse code modulation) has a threshold BER = 10~5 at Eb/N0 = 9.6. Various codes can improve this. The convolutional code with code length 7 and bit rate of half the symbol rate ((7,2) code) is one of the standards -in other words, the system transmits two coded bits for each data bit, with the bits determined by an algorithm with a 'memory' of 7 bits. At a BER = 10~5 it improves link performance by 5.1 dB, i.e. equivalent to a three times increase of the transmitter power or antenna area, although 3 dB of this are spent in the increased bandwidth needed to transmit the two symbols per bit. One common algorithm used to recover data from convolutionally coded data is the Viterbi decoder - its advantage is that it has a fixed decoding time, lending itself to hardware implementation.

Reed-Solomon codes are also common and are particularly robust with respect to short-burst errors - a common code is (255,223) where 32 bytes of parity symbols are added to 223 bytes of data. This allows the correction of up to 16 error bytes. A common combination is first to apply a Reed-Solomon code and then a convolutional coding.

'Turbo codes' (another combination) can give an additional gain of 4 dB but are computationally demanding and require large frames of data (2000-8000 bits) for coding and decoding, which can be risky for the entry probes since possible losses of signal can affect a significant amount of data.

0 0

Post a comment