長期演進計畫系統下網路訊務之識別機制

Abstract

在長期演進技術 (LTE) 中，採用了非連續接收機制 (DRX) 去降低使用者的耗電量。非連續接收機制的循環週期必須被設置能符合網路的訊務以至於達到最佳的省電效應。如果沒有資料要傳送或接收，使用者將會關上傳送或接受器去省電。在本論文中，我們提出了一種能去識別長期演進技術中網路訊務的方法。首先，我們採用了支持向量機 (SVM) 去將網路訊務做分類。接著我們使用簡易突發性數據分布 (bursty packet data distribution) 去模擬背景訊務，以及定義一個能偵測即時訊息訊務的數值。最後我們提出了一種計算方法能識別背景訊務以及即時訊息訊務的訊務特徵。模擬結果顯示出能用來代表線上訊務的數據。當我們不斷得到封包的情況下，我們便不斷的更新這些數據，根據這些數據我們便能識別我們正在使用的訊務狀況以及特徵。這機制可以提供一個方針給非連續接收機制的研究者使用。

Discontinuous reception (DRX) Cycle mechanism is adopted in Long Term Evolution (LTE) system to minimize the power consumption of user equipment (UE). DRX Cycle must be set to suit the Internet traffic to obtain the optimum performance for power saving. If there is no packet to transmit/receive, UE will turn off the transmitter/receiver based on DRX Cycle to reduce the power of UE. In this thesis, we proposed a mechanism to identify Internet traffic in LTE system. First, we adopted Support Vector Machine (SVM) algorithm for traffic classification. Then, we modeled Background traffic with a simplified bursty packet data distribution and defined a value to detect the use of Instant Message traffic. Finally, we proposed a method of calculation to recognize the characteristics of Background traffic and Instant Message traffic. Simulation results show the numerical parameters to represent on-line traffic. With these constantly updating parameters from the packets we captured, we can recognize the situation and characteristics of the traffic we are using. It provides a guideline for DRX researchers.