In this thesis, several design methods for transmit-receive filter pairs and algorithm for their low-cost implementation are proposed.
The transmit-receive filter pair design is nonlinear combinatorial optimization problem, due to its non-linearized ISI constraints, if only discrete values are allowed for the filter coefficients. Therefore, several optimization algorithms, such as local search, mixed integer linear programming (MILP), genetic algorithm, simulated annealing, and dynamic programming are applied for filter pair design problem. The proposed local search method is improved compared with the conventional one and the nonlinear optimization problem is relaxed to allow MILP for suboptimal linear optimization. The other optimization methods applicable to filter design are proposed as well. As an example, the receive filters for WCDMA are designed and the performance and computational load are compared.
The algorithm for obtaining the low-cost structure of transmit-receive filter supporting multiple standards is presented. The transmit-receive filters for different communication system standards are designed separately in the form of FIR filters. For low-cost implementation, the filters should share their hardware resource so that overall cost of the filters could be minimized. Each coefficient of lower order filter is assigned to one of the coefficients of higher order one such that the shared resource saves the hardware cost. Sharing hardware resource is considered as resource assignment problem, which is solved by dynamic programming. An example of multi-mode digital downconverter for the $3^rd$ generation communication systems is considered and approximately 36% reduction of hardware cost is achieved.