快速的動態隨機程序預估演算法：使用平行遞迴類神經網路

A Fast On-Line Training Prediction Technique Pipeline Recurrent Neural Network

沈伯凡(Po-Fan Shen)

《興國學報》 7期 (2007/01) Pp. 173-189

https://doi.org/10.29950/JHKUM.200701.0011

並列摘要

In variant fields of applied science or engineering, control algorithms con be designed with much improved performance if the design factors of the control algorithm can be predicted in advance. In this paper, we investigate the prediction and on-line training capability of pipelined recurrent neural networks (PRNN) by applying it in the wireless CDMA cellular networks. We present a design procedure by applying PRNN to a time correlated process. We also examine the relationship among design factors and the prediction capability. The results show that the prediction error in short-term fading environment is large especially the frame length approaches to the fade duration. This is because the low correlation of the interference process in the short-term fading environment. Therefore, it need further information to design a effective predict based access control.

參考文獻

Liu, Z.,Zarki, M. E.(1994).SIR-based call admission control for DS-CDMA cellular system.IEEE JSAC.12(4),638-644.

Brand, A. E.,Aghvami, A. H.(1996).Performance of a joint CDMA/ PRMA protocol for mixed voice/data transmission for third generation mobile communication.IEEE JSAC.14(9),1698-1707.

Sampath, A.,Holtzman, J. M.(1997).Access control of data in integrated voice/data CDMA system: Benefits and tradeoffs.IEEE JSAC.15(8),1511-1526.

Ishikawa, Y.,Umeda, N.(1997).Capacity and performance of call admission control in cellular CDMA systems.IEEE JSAC.15(8),1627-1635.

Chang, L. F.,Ariyavisitakul, S.(1991).Performance of a CDMA radio communication system with feed-back power control and multipath dispersion.IEEE GLOBECOM'91.(IEEE GLOBECOM'91).