近年 Data Distribution Service (DDS) 逐漸成為管理關鍵任務訊息的趨勢，DDS 作為在應用程式之間以 publish-subscribe 模式傳輸資訊，而我們經常會以 Reliability 與 Throughput 作為傳輸品質標準。DDS 在傳遞資訊時經常會發生一些問題，例如封包遺失或封包延遲送達，兩者分別為Reliability 與 Throughput 問題。為了解決這些問題，DDS 提供了豐富的 QoS 策略能夠用於調教系統傳輸訊息的品質。但是，由於資源的限制，相同的 QoS 策略不能在所有環境種解決所有問題，需要根據資源的限制調整 QoS 策略的配置。 我們為了解決傳輸資料Reliability 與 Throughput 的問題，進行數個情境的實驗。情境一：此實驗的目的是為了瞭解系統中資源的狀態(系統的資源是否充足)。情境二：我們在使用者感興趣的所有主題，試圖調整 QoS 策略，藉由將所有封包暫存在Cache，並重送遺失封包，達到盡可能提升Reliability的效果。情境三：我們只在某些主題試圖調整 QoS 策略，達到盡可能提升Reliability的效果。情境四：我們試圖調整 QoS 策略，藉由將所有封包暫存在Cache，並限制傳送速度，達到盡可能提升Reliability的效果。情境五：我們試圖調整 QoS 策略，藉由增加封包數目的最大值，達到盡可能降低封包的傳送時間，及盡可能提升Throughput的結果。 本研究為提出 QoS 策略配置的演算法，能夠根據系統資源限制的情況下解決傳輸資料Reliability 與 Throughput 的問題 ;Data Distribution Service (DDS) is becoming the most recent trends in net centric system and able to do mission critical information management. DDS is an application which acts as the middleware between applications and use publish-subscribe communication model. In DDS, Data Reliability and Throughput are two important performances quality that can be measured. In many cases, during the communication in DDS there will be some problem occurred, such as packet loss rate and outdated packets which lead to data reliability and throughput problem respectively. In order to deal with those issues, OMG as the standard of DDS, provide rich set of adjustable QoS policy which enable the system to optimize the communication in DDS including the data reliability and throughput. However, the same QoS policy settings cannot be used to solve all the problem due to the resource constraint. The DDS might run under enough resource or limited resource condition. In order to deal with the data reliability and throughput issue over specified resource condition, different QoS policies settings are required. Several experiment scenarios are conducted in order to solve the data reliability and throughput problem. In scenario 1, the experiment will be conducted in order to find the resource state of the system (whether the system is running under enough resource or limited resource). In scenario 2, the QoS policy will be adjusted to optimize the data reliability for all topics by keeping all the packets into cache history, and resend the missing packets. In scenario 3, the QoS policy will be adjusted to optimize the data reliability for several topics only. In Scenario 4, the QoS policy will be adjusted to optimize the data reliability for all topics by keeping all the packets into cache history and limit the sending rate. In scenario 5, the QoS policy will be adjusted to minimize the period of the packets transmission and optimize the throughput by increasing the number of max samples. Based on the conducted experiments, the proposed QoS policies settings can solve data reliability and throughput problem when the system is running under enough resource or limited resource condition.
