<< Chapter < Page | Chapter >> Page > |
In the third design stage, the choices are less constrained. Elements of the third stage are shown inthe lower half of the receiver schematic (the “adaptive layer” of [link] ) and include the selection of algorithms forcarrier, timing, frame synchronization, and equalizer adaptation. There are several issues to consider.
One of the primary stage three activities is algorithm selection—which performance function to use in each block.For example, should the ${\mathcal{M}}^{6}$ receiver use a phase-locked loop, a Costas loop, or a decision-directed method for carrier recovery?Is a dual loop needed to provide adequate carrier tracking, or will a single loop suffice?What performance function should be used for the equalizer? Which algorithm is best for the timing recovery?Is simple correlation suitable to locate the training and marker segment?
Once the specific methods have been chosen, it is necessary to select specific variables and parameterswithin the algorithms. This is a traditional aspectof engineering design that is increasingly dominated by computer-aided design, simulation, and visualization tools.For example, error surfaces and eye diagrams can be used to compare the performance of the various algorithms in particular settings.They can be used to help determine which technique is more effective for the application at hand.
As software-aided design packages proliferate, the need to understand the computational mechanicsunderlying a particular design becomes less of a barrier. For instance, Software Receiver Design has relied exclusively on the filter design algorithms built into M atlab . But the specification of the filter (its shape,cutoff frequencies, computational complexity, and filter length) cannot be left to M atlab . The more esoteric the algorithm, the less transparent is theprocess of selecting design parameters. Thus, Software Receiver Design has devoted considerable space to the design and operation of adaptive elements.
But, even assuming that the tradeoffs associated with each of the individual components are clear, how caneverything be integrated together to succeed at a multifaceted design objective such as the ${\mathcal{M}}^{6}$ receiver?
Even when a receiver is fully operational, it may not decode every symbol precisely. There is alwaysa chance of error. Perhaps part of the error is due to a frequency mismatch, part of the error is due tonoise in the channel, part of the error is due to a nonoptimal timing offset, etc.This section (and the next) suggest a general strategy for allocating “part of” the error to each component.Then, as long as the sum of all the partial errors does not exceed the maximum allowable error, there is a goodchance that the complete receiver will work according to its specifications.
The approach is to choose a method of measuring the amount of error, for instance, the average of the squared recovery error.Each individual component can be assigned a threshold, and its parameters can be adjusted so that it does not contributemore than its share to the total error. Assuming that the accumulation of the errors from various sourcesis additive, the complete receiver will have no larger error than the concatenation of all its parts.This additivity assumption is effectively an assumption that the individual pieces of the system do not interact witheach other. If they do (or when they do), then the threshold allotments may need to be adjusted.
Notification Switch
Would you like to follow the 'Software receiver design' conversation and receive update notifications?