摘要:
A microprocessor having a single common data bus (17) to which the output (33) of the arithmetic-logic unit (11) as well as input and output of the data memory (13) are connected without intermediate buffer registers. Of the working registers (21, 23, 25, 27) connected to the ALU inputs, one group (21, 23) is loaded from the common data bus and the other group (25, 27), used as accumulators, is directly loaded from the ALU output. Specific control circuitry (51, 53, 55, 57, 59, 61) allows selective storing of ALU output values into accumulators (25, 27), and simultaneous transfer with selective scaling into another register and into an addressed memory location within the same cycle during which the instruction was executed.
摘要:
For adaptively adjusting the gain in a modem receiver comprising two amplifiers (15, 19), the following steps are performed: for a buffer (21), a Hilbert filter (23), and an equalizer (27), a respective energy indicator (P.sub.max.sup.2, u.sub.avg.sup.2, x.sub.avg.sup.2) is generated from the signal samples in the respective delay line. Each energy indicator is compared to an associataed upper target level (4L.sub.U, 3L.sub.P, 1.19 L.sub.X), and for the equalizer also to an associated lower target level (0.84 L.sub.X). If necessary, a gain correction factor (S) initially set to 1.0 is modified to obtain an overall gain that keeps delay line energies within desired targets. Target comparisons are made so that excess energy in the buffer or Hilbert filter result in a rapid gain reduction whereas average equilizer energy is used for slow adaptations. After a gain change, all acquired samples in the delay lines are also multiplied by the correction factor (S) so that none is lost during gain acquisition. Gain is distributed between the two amplifiers in a swapping operation without modifying the adjusted overall gain.
摘要:
For rapidly detecting a periodic training signal in a modem receiver, delay line storage (21) is provided for a signal section which is equal to one training signal period (q.multidot.M samples) plus an additional adjacent window (q.multidot.W samples). Thus, the stored signal section includes two windows (17', 19) which are offset by one training signal period and which are similar if a training signal is present. In evaluation means (41) connected to the two windows (17', 19), two auxiliary signals (z.sub.1, z.sub.2) are generated (61, 65) which represent the signal energy and the correlation, respectively, of the samples in the two windows. From the auxiliary signals, a dissimilarity metric (z.sub.3) is derived (69) and is compared (73) to a given threshold for generating an indicator signal (CYC) which becomes active when a training signal is present, i.e. when the two window contents are similar and the metric (z.sub.3) falls below the threshold. A carrier frequency offset estimate (.DELTA..sub.f) is also derived from the second auxiliary signal (z.sub.2).
摘要:
A communication method and arrangement are disclosed which provide, for a trellis encoding process in the transmitter and a maximum-likelihood decoding process in the receiver, an additional fifth state (ZS) in addition to the usual four coding states (S0 . . . S3). The fifth state is entered in the transmitter at the end of transmission when a particularly selected escape symbol from a set of four escape symbols (E0 . . . E3) each associated to one of the four normal coding states is sent by an encoder supplement (15). In the receiver, a stop signal indicating end of transmission is only generated when, after the usual decoder delay, a path decoder (63) determines that an escape symbol was received, and a trellis decoder (61) indicates the fifth state (ZS) to be the most probable one. This enables rapid and secure determination of the end of a transmitted sequence without the need for additional transmission symbols or decoder cycles.
摘要:
A method and apparatus is disclosed to map a sequence of data to Quadrature Amplitude Modulation (QAM) constellation symbols. The method and apparatus encodes only a portion of the sequence of data and leaves a remaining portion of the sequence of data unencoded. The encoded portion of the sequence of data and the remaining unencoded portion of the sequence of data are then mapped into modulation symbols of the QAM constellation. The encoded portion of the sequence of data selects subsets of the QAM constellation, and the remaining unencoded portion of the sequence of data determines a specific modulation symbol within each subset of the QAM constellation.
摘要:
Certain aspects for the start-up procedure of transceivers supporting higher data rates over twisted-pair copper cabling are provided for 10 Gbit/sec Ethernet links (10GBASE-T). During a PMA (physical medium attachment) training period of the start-up procedure, long PMA training frames are exchanged periodically between link partners. A significant portion of each PMA training frame consists of known pseudo random sequences simultaneously transmitted over four wire pairs. PMA training frames include an InfoField for exchanging parameters and control information between link partners. For example, the InfoField's payload comprises fields for indicating current transmit power backoff (PBO), next PBO, requested PBO, transition count, control information, and for communicating precoder coefficients. Information in InfoFields is repeated and is not necessary that a link partner decodes every InfoField. For example, by occasionally reading the transition count, a link partner can determine when a change in transmit PBO and/or a state transition is to occur.
摘要:
In a communication system, Huffman coding techniques are used to obtain shaping gains for an improvement in data transmission rates. More particularly, a novel method of Huffman shaping is described that achieves a shaping gain of greater than 1 dB. The shaping gain results in a higher data rate transmission in a communication system where transmitted power is constrained.
摘要:
A method for modulating a sequence of data symbols such that the transmit signal exhibits spectral redundancy. Null symbols are inserted in the sequence of data symbols such that a specified pattern of K data symbols and N−K null symbols is formed in every period of N symbols in the modulated sequence, N and K being positive integers and K being smaller than N.
摘要:
Cancellation of interference in a communication system with application to S-CDMA. A relatively straight-forward implemented, and computationally efficient approach of selecting a predetermined number of unused codes is used to perform weighted linear combination selectively with each of the input spread signals in a multiple access communication system. If desired, the predetermined number of unused codes is always the same in each implementation. Alternatively, the predetermined number of unused codes are selected from within a reordered code matrix using knowledge that is shared between the two ends of a communication system, such as between the CMs and a CMTS. While the context of an S-CDMA communication system having CMs and a CMTS is used, the solution is generally applicable to any communication system that seeks to cancel narrowband interference. Several embodiments are also described that show the generic applicability of the solution across a wide variety of systems.
摘要:
In the echo-cancelling device of the invention new decision-error directed algorithms are developed which permit to maintain adaptivity of the echo coefficients and perform precise tracking of the far-end echo phase in a computationally efficent manner during full-duplex operation. The echo-cancelling device comprises an adjust device (40) performing the minimization of the mean-square error obtained as the difference between the signal at the output of the equalizer (Z) and the data-symbol decision (a), for providing the adjustment of the near echo estimator (20) coefficients, of the far echo estimator (22) coefficients, and of the phase (26) of the estimated far echo.