C-OFDM for stationary channels

Reduction of the channel model

We see the OFDM method not as a procedure but rather as a solution. A solution to reduce a channel with memory and inter-symbol-interference into a set of parallel, independent channels with no memory and no inter-symbol-interference, without changing the channel capacity.

Further the channel is reduced by shortening the channel impulse response by a patented synthesis of the receiver filter, that again doesn’t affect channel capacity. The channel capacity can be approached by the optimization of signal code construction, which includes bit loading and highly efficient coding.

In many applications we use 4D-Trellis Coding with Forneys Sign-Trellis-Shaping and Viterbi Decoding. If necessary this scheme is supplemented by a special outer Reed Solomon Coding, that is designed for our COFDM structure.

Initial Estimation of the unknown stationary channel

This includes estimation of

  • Signal level and AGC setting
  • Clock and Frequency-Offset and Correction
  • Noise level and noise classification (background noise, narrowband interferers, coloured noise)
  • Channel transfer function

Optimization of OFDM Signal-Code-Construction

We derived a solution to the theoretical optimization problem to reach the maximum data transmission rate in the given channel with a limited signal level at the channel input and a limited bit error probability.

An equivalent optimization problem is to minimize the bit error probability with a given data rate and same signal level restrictions. Our modems are solving this optimization problem during the operation. As a result the bit-loading and coding scheme is adapted to the channel to exploit channel capacity.

Further we optimize the OFDM signal architecture to cope channels with very long impulse responses. Here we find an optimized operating point in terms of minimum delay and maximum performance.

Adaptation of transmitter and receiver

Elements of transmitter and receiver are adapted to the underlying channel. In transmitter we are able to apply sharp-edged filters to limit the spectrum of the transmitted signal and to avoid out-of-band interferences.

Further we apply a special kind of predistortion in the transmitted signal for compensating the channels amplitude response already on transmitter side. This fact in combination with an optimised individual selection of the information density in each subchannel yields the following result: the bit error probability is equal in all subchannels- the best situation for efficient coding.

The receiver input filter is adapted to shorten the channel impulse response avoiding intersymbol-interference. The equalizer is subdivided into two stages which are adapted by Kalman-Filtering.

Reaction to fast and slow variations of the channel characteristic

During operation the modem is adapting to the channel variations and controlling the result by measuring the transmission quality in terms of block error rate and Euclidean distances. If channel conditions change, the modem can react by increasing or decreasing the transmission rate, by changing the OFDM signal code construction and symbol architecture. We also apply procedures that enables the modem to cope even channels with very long impulse responses.

Synchronization

The synchronization is done in two steps. We apply a set of fast algorithms for initial synchronization of clock, frequency and symbol shift. During data transmission the clock and frequency synchronization is preserved by a reliable algorithm with linear regression and Kalman filtering.

Further advantages of our OFDM technique

OFDM Modulation with

  • Software coding of data rate and bandwidth
  • Optimization of signal code construction (adaptation of data rate, signal constellation and coding to the channel quality)
  • Steep skirt filters in transmitter and receiver for limited spectrum
  • Pre-distortion of the transmit signal to minimize bit error probability
  • Multi-finger OFDM with selective allocation of available bandwidth
  • Adaptive Length of Guard Intervall, to cope channels with long impulse response

Continuous optimization of the transmission quality through

  • Adaptive pre-equalization at the receiver with filtered coefficients
  • Receiver equalization with decision feedback
  • Adaptation of the equalizing coefficients with Kalman filtering and cycling redundancy check
  • Adaptation of the receive filter with projection of the gradient
  • Adaptation of frequency and timing control with linear regression
  • Data rate Fallback/fall-forward algorithms