車載網路緊急訊息傳遞之地理樹狀演算法 - 政大學術集成

全文

(1)九十九學年度政大資訊科學系碩士論文. A Geo-based tree algorithm for Emergency message delivery in 802.11p vehicular networks. 車載網路緊急訊息傳遞之地理樹狀演算法. Advisor：Tzu-Chieh Tsai Student： Hsin-Chi Wang 05/04/2011.

(2) 車載網路緊急訊息傳遞之地理樹狀演算法. 摘要. 在現今車載傳輸安全資訊中最重要的兩個評估效能參數 -- Latency 和 reliability，要兼顧到這兩個的效能在現今不可靠的無線連接中是很困難的工作。這主要的挑戰是來自於在網路連接層的廣播，使用了不可靠的傳輸，例如:當某點接受到一個訊息時並不會傳輸一個反饋的信息給來源端通知它已接受到此信息。在現今有許多的方法是利用多於的點去增加接受的可靠性，但這將會增加網路頻寬的負載。在我們的論文中我們提出了一個新的傳輸安全資訊的方法，使用少數 relay 去完成快速的安全資訊傳輸，並且在相同時間內去保持高效能傳輸的可靠性。在這我們介紹了兩個 relay 的型式同時地去減少 end-to-end 傳輸延遲時間和增加傳輸範圍可靠性。地理樹狀演算法是由樹演算法而來的，它可以減少不必要的 relay 和傳輸資訊碰撞的發生，而 802.11p 則是用在優先權比較高的點擁有比較小的重新傳輸延遲時間。. i.

(3) A Geo-based tree algorithm for Emergency message delivery in 802.11p vehicular networks. Abstract. Vehicular environments impose a set of new requirements on today’s wireless communication systems. Latency and reliability are very important means to disseminate safety information like time-sensitive emergency messages (EMs) in Vehicular Ad hoc Networks (VANETs). Providing low-latency, high-coverage and scalable multi-hop EM broadcast is a hard task in VANET with unreliable links. The major challenge comes from the fact that the link-layer broadcast uses unreliable transmissions, i.e., no positive feedback to acknowledge the reception of the message. Many existing works have used redundant relay nodes to enhance the reliability of broadcast packet reception. However they often involve more relays than it is necessary, which increases the network load and undermines the scalability of the protocol. Moreover, large latency is often incurred due to coarse protocol design. In this thesis, we propose a new EM broadcast scheme that uses a small number of relays to achieve fast multi-hop EM propagation, at the same time to maintain a high level of transmission reliability. Two types of relays are introduced to reduce latency and to enhance reliability simultaneously, so that low-latency, the desired reliability level and small overhead can be achieved at. ii.

(4) the same time. The inverse tree algorithm (ITA) is based on tree algorithm, the mechanism to select single relay distributive, which features an effective redundant relay suppressing mechanism and very small rebroadcast delay for high priority nodes. Simulation study shows that ITA and multi-channel achieves close to 100% reliability, while using a small number of relays with very low broadcast latency under a wide range of road traffic conditions.. iii.

(5) Acknowledgements. I would like to gratefully acknowledge the support in this work of many people.. First, I wish to express sincere appreciation to my advisor, Dr.. Tzu-Chieh Tsai, for his constant support.. Without his help, it would be. impossible for me to finish this work.. I would also like to thank Sheng-Zhi Zhuang for his invaluable advice and enriching my knowledge and provide spirit during my study.. Lastly, I would like to thank my family for their unconditional support. I dedicate this thesis to my parents. And also I dedicate this effort to my wife.. iv.

(6) TABLE OF CONTENTS. CHAPTER 1. Introduction, Application and Motivation ............................ 1. 1.1.. Introduction ..............................................................................................................1. 1.2.. Application ...............................................................................................................3. 1.3.. Motivation ................................................................................................................5. CHAPTER 2 2.1.. Related Work............................................................................ 6. Opportunistic Broadcast protocol ---- OBP..............................................................6. 2.1.1. Forwarder .................................................................................................................7 2.1.1.1 How to select ..........................................................................................................7 2.1.2. Makeups ...................................................................................................................8 2.1.2.1 How to select ..........................................................................................................9 2.1.3. Disadvantage ............................................................................................................9. 2.2 Efficient Directional Broadcast ---- EDIB........................................... 10 2.2.1. How to select black-burst period and forwarder ....................................................10 2.2.2 If happen collision .................................................................................................. 11 2.2.3 System model .........................................................................................................12 2.2.4 Disadvantage ..........................................................................................................12. CHAPTER 3. GEO-based Tree Algorithm (GTA) ...................................... 13. 3.1. Overview ................................................................................................................13 3.2. IEEE802.11p multi-channel ...................................................................................16 3.3. Tree Algorithm (TA).............................................................................................. 19. v.

(7) 3.4. GEO-based Tree Algorithm (GTA) ........................................................................21 3.4.1 What need Makeup.................................................................................................27 3.4.2 Follow chart............................................................................................................27. CHAPTER 4. Simulation............................................................................... 28. 4.1. Simulation setup.................................................................................................28 4.2.. Simulation1 – 3X3 street and 6 paths.................................................. 29 4.3. Simulation2 – 6X5 street, 3X2 viaduct and 7 paths for real map ....... 36 4.4. Simulation3 – base on simulation2, but use single channel ................ 43 CHAPTER 5. Conclusion .............................................................................. 46. References .......................................................................................................... 47. vi.

(8) LIST OF TABLES. Table 1: EDCA paramemter set used on the CCH. ............................................. 18 Table 2: Default EDCA parameter set on the SCH............................................. 18 Table 3: Parameter settings ................................................................................. 28. vii.

(9) LIST OF FIGURES Figure1-1: The DSRC spectrum band and channels in the U.S............................ 1 Figure1-2: Distance V.S. Throughput for difference application on the 5.9GHZ 4 Figure2-1: VANET model and overview of the broadcast scheme……………..6 Figure3-1: Tree Algorithm V.S. GEO-based Tree Algorithm ............................. 13 Figure3-2: GEO-based Tree Algorithm system model ....................................... 14 Figure3-3: WAVE Multi-channel Operation ....................................................... 16 Figure3-4: Reference architecture of the MAC with channel coordination ....... 17 Figure3-5: Portion of the WSMP header used to control PHY transmit parameters . ................................................................................................. 18 Figure3-6: Tree Algorithm process. .................................................................... 19 Figure3-7: FCFS splitting algorithm................................................................... 20 Figure3-8: GEO-based Tree Algorithm power range.......................................... 21 Figure4-1: Simulation1-- 3X3 street and 6 paths................................................ 32 Figure4-2: Average packet reception ratio (Reliability) for simulation1............ 33 Figure4-3: Average broadcast latency for simulation1. ...................................... 34 Figure4-4: Average number of relays for simulation1........................................ 35 Figure4-5: Average number of overhead for simulation1................................... 35 Figure4-6: Simulation2-- 6X5 street, 3X2 viaduct and 7 paths.......................... 39 Figure4-7: Simulation2-- 6X5 street, 3X2 viaduct and 7 paths for real map ..... 40 Figure4-8: Average packet reception ratio (Reliability) for simulation2............ 41 Figure4-9: Average broadcast latency for simulation2. ...................................... 41 Figure4-10: Average number of relays for simulation2 ...................................... 42 viii.

(10) Figure4-11: Average number of overhead for simulation2 ................................. 42 Figure4-12: Average packet reception ratio (Reliability) for simulation3.......... 43 Figure4-13: Average broadcast latency for simulation3. .................................... 44 Figure4-14: Average number of relays for simulation3 ...................................... 44 Figure4-15: Average number of overhead for simulation................................... 45. ix.

(11) CHAPTER 1 Introduction, Application and Motivation 1.1. Introduction Communication in Vehicular Ad Hoc Networks (VANETs) has been an active research area in recent years. Enabled by Dedicated Short Range Communication (DSRC) [1], these networks are designed to provide a wide range of applications such as safety warning, congestion avoidance or mobile infotainment. The DSRC spectrum band and channels in the U.S show in Figure1-1 [16]. It frequency range is 5.850G～5.925GHz have 75MHZ bandwidth, use 10MHz division to 7 channels (1 control channel + 6 service channels). One of the most important applications of VANET is the multi-hop broadcast of emergency messages (EMs) like hazard warning. Often, EMs need to be sent onto a long backward and forward road segment to notify as many upcoming vehicles as possible and as soon as possible. This necessitates the use of multi-hop broadcast, which extends the broadcast range to several thousands meters. Moreover, the broadcast service needs to have good performance under different traffic scenarios (dense and sparse network). Therefore, the main goals of EM broadcast are high coverage, low-latency and scalability.. Figure1-1: The DSRC spectrum band and channels in the U.S. 1.

(12) However, in real VANETs these goals are hard to achieve at the same time. The major challenge comes from unreliable wireless links [2, 3], which undermine the reliability of single-hop broadcast, the building block of multi-hop broadcast. According to studies on the existing DSRC [4], the one-hop broadcast reception rate is low. This is because channel fading makes the probability of successful packet reception decrease with distance, and packet collisions could rise from hidden terminals due to the lack of channel resource reservation, which gets even worse in a dense network with congested channel. Unlike unicast, there is no positive feedback to acknowledge the reception of a broadcast message. Therefore, no guarantee of packet reception can be made for a single-hop link layer broadcast. In order to enhance multi-hop broadcast reception rates, most previous works have focused on redundant relay retransmission strategies from network layer. While blind flooding leads to the well-known broadcast storm problem [5] where packet collisions arise due to uncoordinated simultaneous rebroadcasts, various methods were proposed to mitigate this problem, such as probability-based methods [6] and temporal ordered retransmissions [7, 8, 9, 10, 11, 12]. However, they often spend more redundant transmissions than it is necessary which increases channel load and reduces scalability. Moreover, the broadcast latency is often large due to the big time delay needed to separate two subsequent retransmissions to avoid collision. In this paper, we propose a new multi-hop EM broadcast scheme for VANET with unreliable links. Our main contributions are two-fold. First, the ITA scheme obtains a nice balance between reliability and overhead, where the basic component is Tree Algorithm (TA) [14] and while achieving low latency. The TA is a collision resolution interval: split left/right or more side. When traffic high, TA base on TDMA technology to maintain throughput in stable status. In this way, the number of redundant relays is effectively cut. 2.

(13) down, and the rebroadcast delay can be very small. The relays are classified into forwarder which provides fast router-propagation and reduces broadcast latency. Second, we propose an 802.11p [13] multi-channel to actually select those makeup, where makeup which enhances reliability of nodes. We will use 802.11p multi-channels to maintain reliability of nodes, the multi-channel include a control channel (CCH) plus six service channels (SCH). The selection process of makeups is FDD optimized so that the total number of relays is reduced. Avoid packet collisions and storm problem. NS-2 [15] simulations under street scenario show that 802.11p multi-channel and ITA are able to achieve close to 100% EM coverage with very small latency when the network is well connected; and the scalability to the dense or sparse traffic regime in terms of number of relays is good.. 1.2. Application IEEE 802.11p (Wireless Access in Vehicular Environments, WAVE). We explain the relation document as IEEE1609.X series standard below: . 1609.1: Resource Manager. . 1609.2: Security Services for Applications and Management Messages: provide Security encryption for application program and management message.. . 1609.3: Networking Services: provide WAVE address and rout service.. . 1609.4: Multi-channel Operation: control channel、service channel、priority parameter access、channel exchange、route management service and multi-channel select.. We will focus of IEEE1609.2 application. Here have four applications for life to satisfy our request in difference application (Figure1-2). . Basic safety message (we research direction). 3.

(14) . Tolling. . General Internet access. . Roadside e-commerce. Figure1-2: Distance V.S. Throughput for difference application on the 5.9GHZ [20]. IEEE802.11p has two key components: Road Side Unit (RSU) and On Board Unit (OBU). In difference application, need both/any RSU and OBU to connect each other. IEEE 802.11p two feature: compatible others standard and upgrade 802.11a spec for high speed move.. 4.

(15) 1.3. Motivation IEEE802.11p goal support emergency message delivery to reach increase reliability and decrease time delay and overhead. Final, we challenge is avoid traffic congestion.. 5.

(16) CHAPTER 2 Related Work. Here we choose two papers to compare with our method. It includes delay time, reliability, relay and overhead number in chapter 4.. 2.1. Opportunistic Broadcast protocol ---- OBP Use two terms to reduce latency and enhance reliability in onehop-zone (Figure2-1). They use NS-2 simulations under highway scenario. Defect: The performance enhancement under the sparse traffic regime is out of the scope of this paper.. Figure2-1: VANET model and overview of the broadcast scheme. 6.

(17) 2.1.1 Forwarder. Reduce delay latency. All nodes are potential forwarder after receives n new EMs. Every node calculate a rebroadcast time delay, after it quickly sends a short ACK at base rate. to. suppress. other. potential. forwarders. (rebroadcast). to. source.. The. potential forwarder that actually rebroadcasts becomes a forwarder. Balance between △ti and NF .. 2.1.1.1 How to select The shortest time delay will become forwarder in onehop-zone.. Where S is the slot number, ρis traffic density andΔti is time delay. Here have two imports. First, when ρincrease as time delay increase. Second, when forwarder distance increase as time delay decrease.. 7.

(18) Send an EM Send an EM. Broadcast. Calculate Time delay. Calculate Time delay. Send an EM. Calculate Time delay. Send an EM. Calculate Time delay. Timer expires ACK. XD become Forwarder. 2.1.2 Makeups. Enhance reliability. A node located in the onehop-zone and receives a new EM from a forwarder. becomes. a. potential. makeup.. A. potential. makeup. that. actually. rebroadcasts becomes a makeup. Balance between reliability and overhead. The optimal boundary xD (makeup) is the middle point of it (xI ) (onehop-zone), the priority of nodes in sub-zone decrease with their distance to xV .. 8.

(19) 2.1.2.1 How to select. Example:. Send an EM Send an EM Send an EM. Broadcast. Select min PRP in onehop-zone. X3 become makeup for M1 Send an EM. Send an EM. M2 PRP > M1 PRP. X2 become makeup for M2 Terminate makeup selection: X4 PRP>Pth. 2.1.3 Disadvantage In high density: Forwarder: Timer is bigger. It will affect delay time performance. Makeup: PRP is poor. It will affect delay performance.. 9.

(20) 2.2. Efficient Directional Broadcast ---- EDIB. The paper use EDIB technology, the EDIB is base on DIB. The farther the distance, the longer the black-burst (RTB message generates) period.. 2.2.1 How to select black-burst period and forwarder. Where L1 is the length of the candidate’s black-burst period, Tslot is a slot time, Dcurrent is the distance between the sender and the candidate.. Send a RTB. Broadcast. Send a RTB Send a RTB Send a RTB. E’s black burst=0 Send a CTS EM ACK. D become Forwarder. 10.

(21) 2.2.2 If happen collision in the same segment. Send a RTB Send a RTB Send a RTB Send a RTB Send a CTS. Collision. Send a CTS. Wait ACK timeout Wi. Wi. Recalculate D and E black burst D’s black burst =1, E’s black burst=2 Retransmit a RTB Send a CTS EM ACK. E become Forwarder. 11.

(22) 2.2.3 System model. Assume: (1) Communication ranges of the vehicles to be much larger than the width of the road. (2) Whenever a vehicle moves out of the 500-th grid, its position will be reset as the first grid. (3) The moving direction (right) of each vehicle is the same as the direction of emergency message dissemination. The transmission range of each vehicle is 20 grids. (4) The available range of each emergency message dissemination is 100 grids. (5) The low bound Vlow and high bound Vhigh of speed are 70km/hr and 150km/hr, respectively. (6) The data size of each emergency message is 30Kbytes.. 2.2.4 Disadvantage In high density: Sub-segment grows large. It will affect delay time performance.. 12.

(23) CHAPTER 3 GEO-based Tree Algorithm (GTA). 3.1. Overview. Figure3-1: Tree Algorithm V.S. GEO-based Tree Algorithm In this paper, we consider the emergency message broadcast in the street and viaduct scenario. We provide Geo-based tree algorithm to increase reliability and decrease delay time. It method is base on tree algorithm. Figure3-1 shows tree algorithm use time to divide and Geo-based tree algorithm use space to divide when collision happened. Figure3-2 shows the base system model, which are two cross street with one lane in each direction. The VANET consists of vehicles that are all equipped with On Board Units (OBUs) that can communicate with each other. Suppose an accident happens in the center road, where the source vehicle stops and its OBU begins to broadcast Emergency Messages (EMs) toward the Interested Region (IR). We divided the IR into three power levels (PL). The PL is defined as the center road segment of length P1, P2 and P3 (P1>P2>P3, Figure3-8). 13.

(24) in the concentric circles of the source vehicle, and the message propagation direction is opposite and identical to the driving direction. Division power goal is let source to know node’s location in onehop-zone. Under this constraint, we want to reduce broadcast latency and enhance reliability.. Figure3-2: GEO-based Tree Algorithm system model. In this paper, two types of broadcast relays are proposed, i.e., forwarder and makeup. The idea is, first employ one forwarder each hop to relay the EMs in the propagation direction so that the farthermost node can receive EM with the highest priority service channel for 802.11p; however this results in uncovered nodes between the hops due to probabilistic reception. Therefore makeup is selected to fill the uncovered nodes in the space between two forwarders which is termed as IR. We aim at reducing the number of forwarders thereby the broadcast latency, and also minimizing the number of makeup in each IR so that given other service channels except the highest priority service channel is. 14.

(25) reached. The latter provides a balance between reliability and overhead. An overview of the broadcast scheme is shown in Figure3-2. Next we give the definitions. (1) DEFINITION 1 (FORWARDER are Pink nodes). A node in the IR that receives a new EM from the source that use GTA (Figure3-8) to select this nodes and assign the highest priority service channel for 802.11p to relay Ems to decrease time delay . (2) DEFINITION 2 (MAKEUP are Green nodes). A node located in the P2. If haven’t any node in P2, It will selected in P1. Makeup receives a new EM from a secure that uses other service channels except the highest service channel to propagation EMs to other nodes in onehop-zone to increase reliability and decrease relay number/overhead.. For each type of relay, the broadcast process it is involved in is called forwarder or makeup phase respectively. There are more multiple routing phases, which terminates until the last forwarder completes its job; and there may be multiple makeup phases, each of which consists of makeup in the closest source power range. The forward and makeup phases are done in parallel, because they use different service channels in the propagation direction broadcasts. Thus, the broadcast latency is reduced. When all nodes receive EMs in IR, they will check WSMP header status. If old node data isn’t empty for WSMP header (Figure3-5), new data compare with old data. If the same, it can’t update new EMs into its WSMP header in IR, on the other hand. It can reduce overhead and memory size.. Next we give several assumptions made in this paper. (1) Vehicles are GPS-capable. Each vehicle obtains its location and speed in real-time. This is widely accepted assumption in VANET literature. (2) Vehicles are aware of the existence and locations of nearby vehicles, as they broadcast. 15.

(26) beacon messages for every △ t = 100ms. This is realistic in the real VANET environment, since for the sake of safety each vehicle must know its distance to others to prevent collision. And it is also required by the WAVE standard [1].. 3.2. IEEE802.11p multi-channel Reference to IEEE_Std_1609.4 [13]: IEEE Trial-Use Standard for Wireless Access in Vehicular Environments (WAVE)—Multi-channel Operation (Figure3-3). The multi-channel includes one control channel and six service channels total 7 channels. The purpose is to enable effective mechanisms that control the operation of upper layer across multiple channels, without requiring knowledge of PHY parameters, and describe the multi-channel operation channel routing and switching for different scenarios.. Figure3-3: WAVE Multi-channel Operation. When the WSMP data is passed from the LLC to the MAC, the MAC shall route the packet to a proper buffer (queue) corresponding to the channel number contained in the. 16.

(27) WSMP header (Figure3-5).. Figure3-4: Reference architecture of the MAC with channel coordination. When IP data (Figure3-5) is passed from the LLC to the MAC, the MAC shall route the packet to a data buffer that corresponds to the current SCH. Highest priority has the smallest Contention Window (CW). CW use random back-off mechanism.. 17.

(28) Figure3-5: Portion of the WSMP header used to control PHY transmit parameters. Set EDCA parameters in CCH (table 1) and SCH (table 2) priority.. 18.

(29) 3.3. Tree Algorithm (TA) Tree algorithm is base on TDMA technology to avoid packets collision. If happen collision, it will use pure binary division time to left and right side in next time. If left or right side happen collision again, it will use pure binary division time to left-left/ left-right/right-left or right-right side in next time. Continue until idle or success is happened.. Figure3-6: Tree Algorithm process. 19.

(30) Figure3-7: FCFS splitting algorithm. Tree Algorithm process has three statuses: (1). If feedback is collision and σ (k) = L T (k+1) = T (k) α (k+1) = ½[α (k)] σ (k+1) = L. (2). If feedback is successful and σ (k) = L T (k+1) = T (k) + α (k) α (k+1) = α (k) σ (k+1) = R. 20.

(31) (3). If feedback is idle, and σ (k) = L T (k+1) = T(k) + α(k) α (k+1) = ½[α (k)] σ (k+1) = L. Masseyo` Improvement. 3.4. GEO-based Tree Algorithm (GTA). Power range: P1>P2>P3 S represent source (event happen). Figure3-8: GEO-based Tree Algorithm power range. First we define further distance between source and forwarder or forwarder and forwarder in next time slot. Forwarder and Makeup selection use GTA and 802.11p multi-channel + CSMA/CA respectively. Here have five scenarios to cover all status.. 21.

(32) Scenario1: Forwarder is success in P1 Makeup is success in P2. Meter. Send P3 command ACK Send P2 command ACK Send P1 command ACK EM. C become Forwarder EM. B become Makeup EM. T. 22.

(33) Scenario2: Forwarder is success in P1 Makeup is collision in P2. Meter. Send P3 command ACK. Broadcast. Send P2 command Send P2 command ACK. Collision. ACK Send P1 command ACK EM. D become Forwarder RTS RTS. CSMA/CA CTS EM. B become Makeup EM. EM. T. 23.

(34) Scenario3: Forwarder is success in P1 Makeup is idle in P2 and success in P3. Meter. Send P3 command ACK Send P2 command Can’t any ACK from P2 nodes Send P1 command ACK EM. B become Forwarder EM. A become Makeup T. 24.

(35) Scenario4: Forwarder is collision in P1 Makeup is success in P2. Meter. Send P3 command ACK Send P2 command ACK Send P1 command Send P1 command. Broadcast. Send P1 command Send P1 command Send P1 command ACK ACK ACK. Collision. ACK ACK. Send P1-L command. P1-L. Send P1-L command ACK. Collision. ACK. P1-LL Send P1-LL command ACK EM. C become Forwarder EM. B become Forwarder. 25.

(36) Scenario5: Forwarder is idle in P1 and success in P2 Makeup is collision in P3. Meter. Send P3 command Send P3 command ACK. Collision. ACK Send P2 command ACK Send P1 command Can’t any ACK from P1 nodes EM. C become Forwarder RTS RTS. CSMA/CA CTS EM. A become Makeup EM. T. 26.

(37) 3.4.1 What need Makeup When source happen traffic accident, it broadcast EMs to all nodes in difference power level. If haven’t Makeup machine, this will happened two status. First, all nodes response ACK to source, this will happened collision. While all nodes product CW to send ACK to source. If CW the same, it will happen collision again. Increase collision as increase CW and increase delay time. Second, all nodes haven’t response ACK to source. This will cause low reliability the same relate work 2.. 3.4.2 Follow chart. 27.

(38) CHAPTER 4 Simulation 4.1 Simulation setup. We evaluate the performance of GTA, and compare it to OBP [17] and EDIB [18]. We use the network simulator NS-2.33 [15], which supports the probabilistic propagation model and enhanced 802.11 MAC layers [19]. The parameters are summarized in Table 3. The other PHY and MAC layer parameters follow the default settings of IEEE 802.11p. Each vehicle generates beacons at a rate of 10 packets/second forroutine safety applications. Event-driven EMs are generated at one vehicle located at x=500m and y=500m and x=750m and y=750m in the street road and x=300m and y=600m in viaduct every 100ms from 100s to 150s. The simulation lasts for 200s.. 28.

(39) We have three cases to simulate in our and other method. Simulation1: 3X3 street and 6 paths (CC=1, SC=2). Simulation2: 6X5 street, 3X2 viaduct and 7 paths for real map (CC=1, SC=2). Simulation3: base on simulation2, but use single channel (CC=1, SC=1). CC is control channel, SC is service channel. 4.2 Simulation1 – 3X3 street and 6 paths Path1 (yellow cars) 00-10-11-12-22-21-20-10-00-01-11-21-22-12-02-01-00. 29.

(40) Path2 (red cars) 01-11-21-22-12-02-01-00-10-11-12-22-21-20-10-00-01. Add Path3 (pink cars) 22-21-20-10-11-12-02-01-11-21-22-12-11-01-00-10-20. 30.

(41) Add Path4 (blue cars) 20-21-11-01-02-12-11-10-00-01-11-21-22-12-02-01-00-10-20. Add Path5 (green cars) 10-11-01-00-10-20-21-11-12-02-01-11-21-22-12-11. 31.

(42) Add Path6 (white cars) 12-02-01-00-10-11-01-02-12-22-21-11-10-20-21-22-12-11. Figure4-1: Simulation1-- 3X3 street and 6 paths. 32.

(43) 4.2.1 Results. Simulation1 is a simple scenario. Six different paths to enhance traffic flow. Figure4-2 to Figure4-5 shows the mean and confidence interval of data points from 100 EMs. A random topology is generated for each road traffic density.. 4.2.1.1 Reliability and Broadcast Latency. From Figure4-2 and Figure4-3 we can see the major different place is EDIB technology. In EDIB, reliability decreases as collision and BER increases in high density. And EDIB hasn’t ACK mechanism to source, besides Forwarder. If there are more vehicles on the road, there exist more candidates during candidate competition phase, this will effect time delay larger than other method. .. Figure4-2: Average packet reception ratio (Reliability) for simulation1. 33.

(44) Figure4-3: Average broadcast latency for simulation1. 4.2.1.2. Relays Number and Overhead. From Figure4-4 and Figure4-5 in high density, OBP is worst method. Because a low PRP cause layer increase. Relay number will grow quickly. EDIB is best method, because it is only forwarder cause reliability is worst. Relays enhance increase as overhead increase.. 34.

(45) Figure4-4: Average number of relays for simulation1. Figure4-5: Average number of overhead for simulation1. 35.

(46) 4.3 Simulation2 – 6X5 street, 3X2 viaduct and 7 paths for real map. Path1 (viaduct--yellow cars) 0-1, 1-2, 2-3, 3-0. Path2 (viaduct--yellow cars) 4-5, 5-6, 6-7, 7-4. 36.

(47) Path3 (street--red cars) 00-000-10-101-11-111-112-12-120-121-22-211-21-201-20-102-100-1001-110-1101 -1102-113-1011. Path4 (street--pink cars) 22-121-120-12-02-012-0102-112-111-0101-011-01-010-11-110-1001-100-102-20-2011011-113-21. 37.

(48) Path5 (street--blue cars) 02-12-120-121-22-211-1102-1101-112-0102-012-011-0101-111-11-110-1101-110-1001-10 0-10-101-11-010-000-00-01. Path6 (street--green cars) 20-201-21-211-22-121-120-1101-110-1001-1011-113-110-11-010-01011-0101-010-000-10. 38.

(49) Path7 (street--white cars) 121-1102-113-1011-102-100-1001-110-1101-120-121-22-211-1102-1101-110-11321-201-20-102. Figure4-6: Simulation2-- 6X5 street, 3X2 viaduct and 7 paths. 39.

(50) Figure4-7: Simulation-- 6X5 street, 3X2 viaduct and 7 paths for real map. 4.3.1 Results. In the simulation4, we use a real and complex street in Taipei city (Figure4-7). It major street matrix is 6X5 street, 3X2 viaduct and 7 paths. It includes some sub-road. Figure4-8 and Figure4-9 shows the mean and confidence interval of data points from 100 EMs. A random topology is generated for each road traffic density.. 4.3.1.1 Reliability and Broadcast Latency From Figure4-8 and Figure4-9 we can see result that simulation2 the same. 40.

(51) simulation1.. Figure4-8: Average packet reception ratio (Reliability) for simulation2. Figure4-9: Average broadcast latency for simulation2. 41.

(52) 4.3.1.2. Relays Number and Overhead. Figure4-10: Average number of relays for simulation2. Figure4-11: Average number of overhead for simulation2. 42.

(53) 4.4. Simulation3 –base on simulation2, but use single channel. 4.4.1 Results Simulation3 the same simulation2, besides reduction service channel from 6 to 1. Figure4-12 to Figure4-15 show the mean and confidence interval of data points from 100 EMs. A random topology is generated for each road traffic density.. 4.4.1.1 Reliability and Broadcast Latency. Figure4-12: Average packet reception ratio (Reliability) for simulation3. 43.

(54) Figure4-13: Average broadcast latency for simulation3. 4.4.1.2 Relays Number and Overhead. Figure4-14: Average number of relays for simulation3. 44.

(55) Figure4-15: Average number of overhead for simulation3. 45.

(56) CHAPTER5 Conclusion. We use two simple relay types to propagation EMs. First, forwarder utilizes Geo-based Tree Algorithm (GTA) and 802.11p SCH1 to reduce collision and delay time latency. Second, makeup utilize 802.11p multi-channel SCH2 to enhance reliability. We define SCH1 use highest priority to propagation EMs. CCH will control that all node (forwarder and makeup) have synchronous time slot before send SCH. Result we prove latency and reliability better then OBP and EDIB for high density condition in the street scenario... 46.

(57) References [1]. DSRC and IEEE 1609 standard family, http://www.standards.its.dot.gov/Documents/advisories/dsrc_advisory.htm.. [2]. M. Torrent-Moreno, M. Killat, and H. Hartenstein. The challenges of robust inter-vehicle communications. IEEE VTC 2005, 1:319–323, 28-25 Sept., 2005.. [3]. M. Torrent-Moreno, F. Schmidt-Eisenlohr, H. Fussler, and H. Hartenstein. Effects of a realistic channel model on packet forwarding in vehicular ad hoc networks. IEEE WCNC, 1:385–391, 2006.. [4]. M. Torrent-Moreno, D. Jiang, and H. Hartenstein. Broadcast reception rates and effects of priority access in 802.11-based vehicular ad-hoc networks. In Proceedings of VANET ’04, pages 10–18. ACM, 2004.. [5]. S.-Y. Ni, Y.-C. Tseng, Y.-S. Chen, and J.-P. Sheu. The broadcast storm problem in a mobile ad hoc network. In IEEE/ACM MobiCom, pages 151–162, 1999.. [6]. N. Wisitpongphan, O. Tonguz, J. Parikh, P. Mudalige, F. Bai, and V. Sadekar. Broadcast storm mitigation techniques in vehicular ad hoc networks. IEEE Wireless Communications, 14(6):84–94, December 2007.. [7]. S. Oh, J. Kang, and M. Gruteser. Location-based flooding techniques for vehicular emergency messaging. In IEEE MobiQuitous, pages 1–9, July 2006.. [8]. O. Tonguz, N. Wisitpongphan, F. Bait, P. Mudaliget, and V. Sadekart. Broadcasting in vanet. In MoVE, pages 7–12, May 2007.. [9]. O. Tonguz, N. Wisitpongphan, J. Parikh, F. Bai, P. Mudalige, and V. Sadekar. On the broadcast storm problem in ad hoc wireless networks. In BROADNETS, pages 1–11, Oct. 2006.. 47.

(58) [10] R. Mangharam, R. Rajkumar, M. Hamilton, P. Mudalige, and F. Bai. Bounded-latency alerts in vehicular networks. In MoVE, pages55–60, May 2007. [11] M. T-Moreno. Inter-vehicle communications: Assessing information dissemination under safety constraints. In WONS, pages 59–64. IEEE, January 2007. [12] H. Alshaer and E. Horlait. Optimized adaptive broadcast scheme for inter-vehicle communication. In IEEE VTC, pages 2840–2844, May 2005. [13] IEEE_Std_1609[1].4_2006. [14] Tree algorithm. [15] Ns2. http://www.isi.edu/nsnam/ns. [16] Daniel Jiang, Luca Delgrossi Mercedes-Benz Research & Development North America, Inc. IEEE 802.11p: Towards an International Standard for Wireless Access in Vehicular Environments. Pages2036-2040, IEEE, 2008 [17] Ming Li and Wenjing Lou. Opportunistic broadcast of emergency messages in vehicular ad hoc networks with unreliable links. July 2008. [18] Shun-Ren Yang and Wei-Hsiang Hung. Efﬁcient Selection of Relay Vehicles for Broadcasting on Vehicular Ad-hoc NETworks. The International Conference on Mobile Technology, Applications & Systems 2008 (Mobility Conference), 10-12 September, 2008. [19] Q. Chen, F. Schmidt-Eisenlohr, D. Jiang, M. Torrent-Moreno, L. Delgrossi, and H. Hartenstein. Overhaul of ieee 802.11 modeling and simulation in ns-2. In MSWiM ’07, pages 159–168. ACM, 2007. [20]. 新通訊 2009 年 7 月號 101 期《技術前瞻》. 48.

(59)