[[alternative]]水下聲波感測網路下探討傳播延遲造成之影響及其解決策略

博士[[abstract]]隨著無線通訊技術的發展與進步，無線感測網路(Wireless Sensor Networks, WSNs)也從陸地延伸至水中。在水中，無線感測網路被稱為水下聲波感測網路(Underwater Acoustic Sensor Networks, UASNs)，使用聲波進行資料的傳輸，然而，水下聲波的傳輸與無線電波的傳輸有許多差異，例如：高傳播延遲、有限的頻寬與低資料傳送速率等等。其中，又以高傳播延遲對整體水下聲波感測網路的影響最為嚴重。 媒介存取控制(Media Access Control, MAC)協定是無線網路中重要的研究議題之一，雖然在無線網路中媒介存取控制協定已被廣泛的探討與研究，然而若將陸地上無線網路存在的協定直接應用在水下聲波感測網路中，也無法順利地運行。這是由於聲波傳輸造成的高傳播延遲，使得媒介存取控制協定無法避免碰撞。因此，有許多專門為水下聲波感測網路設計的媒介存取控制協定，解決水下因高傳播延遲造成的碰撞。然而，這些協定大多使用等待的方式來避免傳播延遲造成的碰撞，雖然避免了資料傳輸的碰撞，卻使得整體網路的頻道利用率與效能降低。 因此，在本篇論文中，我們針對感測網路中熱門的分時多重存取技術(Time division multiple access, TDMA)與四向交握機制(Four-way handshaking mechanism)兩種方式，分別設計下列適用於水下聲波感測網路的媒介存取控制協定： (1) 在DSS-TDMA (Dynamic Slot Scheduling TDMA-based MAC protocol)中，我們利用傳播延遲可以平行傳輸的特性為考量，設計一基於分時多重存取的媒介存取控制協定，不但可以依照資料量動態調整排程的結果，也利用平行傳輸使得整體的頻道利用率提升。 (2) 在CS-MAC (Channel Stealing MAC protocol)中，我們使用等待的四向交握機制來解決隱藏節點的問題(Hidden terminal problem)，然而聲波高傳播延遲的特性使得頻道利用率低落，在此方法中，本篇論文提出一個機制使得控制封包交換造成的頻道浪費得以被使用，並減緩暴露節點問題(Exposed terminal problem)。 (3) 在TLPC (Two-Level Power Control MAC protocol)中，我們提出一個媒介存取控制協定，透過電量控制與非等待式的四向交握機制來解決碰撞的問題。在本篇論文中，我們解決兩個碰撞問題，分別是控制封包與資料封包造成的碰撞問題，稱之為Control/DATA Collision (CDC)。以及過大干擾範圍所造成的碰撞問題，稱之為Large Interference Range Collision (LIRC)。同時，透過電量控制，使得有更多的節點能夠同時進行傳輸，使得頻道利用率得以提升。 模擬的結果顯示，本篇論文提出的三種方法，不但能夠避免水下聲波傳輸造成的碰撞，也能夠提升頻道的利用率等等。[[abstract]]With the development and advancement in radio frequency technology, wireless sensor networks (WSNs) have been extended from land to underwater. In the water, WSNs are commonly known as underwater acoustic sensor networks (UASNs). Acoustic signal is used for data transmission in UASNs. However, there are many differences in the transmission between WSNs and UASNs, such as long propagation delay, limited bandwidth, and low data rate, etc. Among which, the most impact in UASNs is long propagation delay. The design of media access control (MAC) protocol is a topic in wireless networks. Although MAC protocols in terrestrial wireless networks have become broadly studied, the existing protocols cannot be directly applied to UASNs due to the long propagation delay. Long propagation delay causes the collisions in transmission. Therefore, there are many protocols designed to avoid collisions caused by long propagation delay for UASNs. Although these protocols can avoid collisions by deferring a period of time, but their performances in channel utilization are low. Therefore, this dissertation focuses on two popular technologies in UASNs, termed time division multiple access (TDMA) and four-way handshaking mechanism, and designs the following MAC protocols to suit to UASNs. (1) In DSS-TDMA (Dynamic Slot Scheduling TDMA-based MAC protocol), the concurrent transmission is taken into consideration. The proposed protocol can not only adapt the schedule dynamically but also use concurrent transmissions to improve the channel utilization. (2) In CS-MAC (Channel Stealing MAC protocol), a four-way handshaking mechanism with a deferring time period is used to solve the hidden terminal problem. However, the deferring time period causes the low channel utilization. Therefore, in CS-MAC, a strategy is proposed to utilize the waste channel resource and mitigate the exposed terminal problem. (3) In TLPC (Two-Level Power Control MAC protocol), a four-way handshaking MAC protocol with power control is proposed to avoid collision problems. In TLPC, two collision problems are studied, termed Control/DATA Collision (CDC) and Large Interference Range Collision (LIRC). By power control, TLPC can not only avoid CDC and LIRC problems but also improve the channel utilization. Simulation results show that the proposed protocols can not only avoid collisions but also improve the channel utilization.[[tableofcontents]]Contents 1 Introduction 1 1.1 Motivation 2 1.2 Organization 4 2 DSS-TDMA: A Dynamic Slot Scheduling TDMA-based MAC Protocol for Underwater Acoustic Sensor Networks 5 2.1 Problem Statement 5 2.2 Related Work 6 2.3 Network Model 8 2.4 The Dynamic Slot Scheduling TDMA-based MAC Protocol (DSS-TDMA) 10 2.4.1 Slot Selection Constraints 10 2.4.2 The dynamic slot schedule strategies 12 2.5 Simulation Results 20 2.5.1 Simulation Setups 20 2.5.2 Simulation Results 21 2.6 Summary 26 3 CS-MAC: A Channel Stealing MAC Protocol for Improving Bandwidth Utilization in Underwater Acoustic Sensor Networks 27 3.1 Problem Statement 27 3.2 Related Work 28 3.3 Preliminaries 30 3.3.1 Network Model and Assumptions 31 3.3.2 Symbol Definitions 31 3.3.3 Problem statement 32 3.4 The Channel Stealing MAC Protocol (CS-MAC) 34 3.4.1 Overview of CS-MAC 34 3.4.2 Time division for collision avoidance 35 3.4.3 Channel stealing mechanism 37 3.5 Simulation Results 41 3.6 Summary 45 4 A Two-Level Power Control MAC Protocol for Collision Avoidance in Underwater Acoustic Networks 47 4.1 Problem Statement 47 4.1.1 Control/DATA Collision (CDC) Problem 47 4.1.2 Underwater Large Interference Range Collision (ULIRC) Problem 50 4.2 Related Work 52 4.3 Derivation of Interference Range in UWANs 54 4.4 The Two-Level Power Control MAC Protocol (TLPC) 57 4.4.1 Basic Idea 58 4.4.2 Power Estimation 59 4.4.3 Transmission Range Estimation 60 4.4.4 The TLPC MAC Protocol 61 4.5 Simulation Results 63 4.5.1 The gird topology 64 4.5.2 The random topology 70 4.6 Summary 74 5 Conclusions 76 5.1 Contributions 76 5.2 Future work 77 Bibliography 79 Publication List of Chan-Ying Lien 91 List of Figures Figure 2.1 The frame structure in DSS-TDMA. 8 Figure 2.2 The network architecture in DSS-TDMA composes of vast underwater sensors and several sinks. Sensors cooperate with each other for detecting or monitoring task. The sensing data also can be either directly or multi-hop forwarded to the sink. 9 Figure 2.3 Slot Constraints. x is a UT slot, y is a UB slot, and z is a UR slot. 11 Figure 2.4 Two kinds of transmission pairs are in the network: (a) The transmission pair without UB slots; (b) The transmission pair with UB slots. 12 Figure 2.5 s1s3 has been transmitting in slot21, and then the states of all the slots are shown in (a). However, slot32 in (b) is a UB slot between s1s3 and s3s1. Therefore, s1s3 and s3s1 should be classified into a group to schedule. 13 Figure 2.6 The benefit of grouping. The UT and UR slots can be reused easily. 14 Figure 2.7 An example scheduling and shifting policy. 16 Figure 2.8 (a) The scheduling results of Group-I; (b) The scheduling results of Group-II. 17 Figure 2.9 The scheduling results of Group-I and Group-II merge into a frame. 17 Figure 2.10 The combination of scheduling results of Group-I and Group-II when traffic loads are high. 18 Figure 2.11 Channel utilization of various protocols for different number of nodes. 22 Figure 2.12 Network throughput generated by various protocols for different offered load in 8 nodes. 23 Figure 2.13 Average packet delay caused by various protocols for different number of nodes. 24 Figure 2.14 One cycle time caused by various protocols for different number of nodes. 25 Figure 3.1 The concept of PCAP mechanism [20]. 28 Figure 3.2 Comparisons of channel utilization among different data rate in terms of data packet size. 33 Figure 3.3 The concept of CS-MAC. 35 Figure 3.4 An example of the idle waiting time division. The time division (a)from back to front; (b)from front to back. 36 Figure 3.5 Three regions are divided in the network. 38 Figure 3.6 The interference at the receiver does not be taken into consideration. 39 Figure 3.7 The interference at the receiver is taken into consideration. 40 Figure 3.8 Comparisons of the network throughput of four protocols in terms of offered load varied from 0.01 to 0.4 packets/s. 43 Figure 3.9 Comparisons of the channel utilization of four protocols in terms of number of nodes varied from 5 to 20 stations. 44 Figure 3.10 Comparisons of the average delay of four protocols in terms of offered load varied from 0.01 to 0.4 packets/s. 45 Figure 4.1 Scenarios of the CDC problem. 49 Figure 4.2 The effectiveness of the CDC problem. 50 Figure 4.3 The effectiveness of the CDC problem in terms of DSR. 51 Figure 4.4 An illustration of the LIRC problem. 52 Figure 4.5 The attenuation and absorption of acoustic wave in terms of DSR. 55 Figure 4.6 An illustration of the interference range in terms of DSR. 58 Figure 4.7 An illustration of TLPC protocol in time domain. 62 Figure 4.8 An illustration of TLPC protocol in space domain. 63 Figure 4.9 The grid topology. 65 Figure 4.10 The comparison of TLPC, APCAP, and Slotted FAMA in terms of number of collisions per STA. (Offered load=0.8) 66 Figure 4.11 The comparison of number of collisions of TLPC, APCAP, Slotted FAMA, and IEEE 802.11 DCF in terms of different traffic loads. 67 Figure 4.12 The comparison of network throughput of TLPC, APCAP, Slotted FAMA, and IEEE 802.11 DCF in terms of different traffic loads. 68 Figure 4.13 The comparison of energy consumption of TLPC, APCAP, Slotted FAMA, and IEEE 802.11 DCF in terms of different traffic loads. 69 Figure 4.14 The comparison of energy consumption per bit of TLPC, APCAP, Slotted FAMA, and IEEE 802.11 DCF in terms of different traffic loads. 70 Figure 4.15 The comparison of network throughput of TLPC, APCAP, Slotted FAMA, and IEEE 802.11 DCF in terms of different traffic loads. 71 Figure 4.16 The comparison of number of collisions of TLPC, APCAP, Slotted FAMA, and IEEE 802.11 DCF in terms of different traffic loads. 72 Figure 4.17 The comparison of energy consumption of TLPC, APCAP, Slotted FAMA, and IEEE 802.11 DCF in terms of different traffic loads. 73 Figure 4.18 The comparison of energy consumption per bit of TLPC, APCAP, Slotted FAMA, and IEEE 802.11 DCF in terms of different traffic loads. 74 List of Tables Table 2.1 The scheduling order of transmission pairs in Group-I. 19 Table 2.2 The scheduling order of transmission pairs in Group-II. 20 Table 2.3 Simulation parameters of DSS-TDMA. 21 Table 3.1 Symbol definitions. 32 Table 3.2 Simulation parameters of CS-MAC. 42 Table 4.1 Simulation parameters of TLPC. 64[[note]]學號: 898410112, 學年度: 10