Tatsuo Hiramatsu

A reconfigurable processor based on ALU array architecture with limitation on the interconnection

Dynamic reconfigurable processor based on ALU array architecture for consumer appliances is described. To implement reconfigurable system on portable or mobile products, we have tried to develop smaller and powerful reconfigurable processor. We have proposed the ALU array architecture with the limitation on the interconnection for area reduction. By the proposed architecture, we could reduce gate size by 63% on interconnections. Also, we have shown that the performance of proposed architecture is almost the same as one without limitations. The proposed processor is a parallel processing processor that consists of a sequencer and an ALU array, adopted multi threading technology. We have developed compilation tools from source codes written in C language for the proposed processor. We demonstrate the software model of the processor using MPEG-4 video decoding application.

Code tracking loop for M-ary/SS systems using outputs gating of correlators

Code tracking loop for M-ary/SS systems using outputs gating of correlators(GCL) method is proposed for the code tracking of a communication system based on M-ary/SS scheme in which one of several sequences prepared on the tranmitter side depending on several bits of data. In the correlator outputs gating method, a certain threshold level is assumed and all correlatorspsila outputs are blocked when all correlatorspsila outputs are smaller than the threshold level. In the code tracking system using the decision feedback loop(DFL) method that is the conventional method, a deterioration in the characteristic happens when decoding error occurs, that is noise level is high or low SNR. In the correlator outputs gating method, however, all correlatorspsila outputs arenpsilat fed into the synchronous system in a situation with low SNR and a high error rate. So that we can expect the improvement of code tracking characteristic that is often reprented by code tracking error. In this paper, numerical calculation is conducted by means of induced formula. In the numerical calculation, it was found that the code tracking characteristic is improved from conventional DFL system. As a result, we can get stable code tracking. The M-ary/SS scheme is expected mass data transfer for audio-visual(AV) systems. By using the GCL system, we can realize stable mass data transfer for AV systems.

Reconfigurable processor based on ALU array architecture for software radio

We have developed an original reconfigurable processor LSI for software radio systems for consumer products. The die size of the LSI is 5.65 mm by 5.65 mm. The processor is based on ALU array architecture and has original limitations on connections of ALUs. The circuit size is small because of the limitations. It achieves high processing performance by processing multiple threads simultaneously. We have developed a prototype of a broadcasting receiver with the LSIs. The prototype has realised the real-time reception of three kinds of broadcasts by changing software using a maximum of two LSIs. We also have estimated a power consumption of the LSI and have confirmed that power consumption can be reduced by approximately 55% of that of the LSI. By doing so, we have advanced a realisation of software radio for consumer products.

SNR estimation of SS communication using orthogonal code for fading channels

Spread spectrum communication has been used in many fields. In this paper, we propose novel SNR (signal to noise ratio) estimation of SS (spread specyrum) communication for fading channels. In DS(direct sequence)/SS communication, a signal is spread by using a spreading code. In the method, we use sequences (a sequence) that are (is) orthogonal to the spreading code. If we use plural sequences, these sequences must be orthogonal each other. We calculate correlations (correlation) between the received signal and the sequences (sequence). Thus, we can estimate a noise power independently. Using the estimated noise power, we can estimate SNR regardless of kinds of fading. Moreover we donpsilat need to use known pilot symbol. We compare the new method with a conventional method by simulation. In this simulation, we also studied the two cases; one orthogonal sequence case and two orthogonal sequences case. As a result, we show the new method can estimate SNR more precisely than the conventional method. Moreover, two orthogonal sequences case could estimate more precisely than one orthogonal sequence case.