Td-scdma System, Joint Detection Algorithm And Its Multi-core Dsp Implementation

TD-SCDMA system uses some advantage technology, such as the joint detection, smart antenna and UL synchronization, and it combines the advantage of TDD and CDMA. So it is the best one in frequency using field. It's very important to research TD-SCDMA system and the relative technology, now.TD-SCDMA is an interfere-limited system. Here includes multi-path interference, multi-user interference of intra-cell and inter-cell interference. Those destroy the orthogonality of each channel which reduces the utility of frequency-spectrum in TD-SCDMA system. JD takes all user interfereas as useful signals which can make full use of the spread code, range, timing and delay of all users. Thus the multi-path and multi-access interference (MAI) are depressed on a large scale. The core idea of this technology is using equalizing technology to cancel the inter-symbol interference (ISI) and the MAI at the same time. Based on this, we work over the application of JD in TD-SCDMA.In this paper, we introduce the basic formations of the physical channel first of all. And make a discrete-time model and expression with matrix according to the character of TD-SCDMA system. It introduces the principle of JD algorithm in general, and talks about the linear JD algorithm based on the rule of zero-forces (ZF) and minimum mean square error (MMSE) in detail. And then make a comparing of these two rules. The aim of JD is to estimate the signal of users, and the bottle-neck is the huge complexity of computing the matrix inverse operation. So it's important to find a simple and effective way to reduce the complexity. This paper deduces BLOCK-FFT algorithm in particular which is one of the fast matrix inverse methods, firstly, handle system matrix to block right circulation by block extend, then solve the system matrix in frequency. To some extent, it reduces the operatin complexity. Then it finishs the simulation of BLOCK-FFT algorithm, verifies the performance of it.Besides this, this paper study the project implementation of BLOCK-FFT algorithm using multi-core DSP. After the fix-point simulation, it implements the algorithm according to the simulation and finishes the parallel operation on the multi-core DSP system. And then, it finishes the test and verification of the multi-core DSP implementation system. At last, this paper make a brief conclusion and expectation of this study field.