IEEE 802.11p protocol - Related works - WAVE/DSRC電子收費系統在預約區之分析與驗證

Chapter 2 Related works

2.3 IEEE 802.11p protocol

The IEEE [6][7][8][9][10][11] makes use of the PHY supplement 802.11a and the MAC layer QoS amendment from 802.11e. The Orthogonal frequency-division multiplexing (OFDM) utilized in the 802.11p PHY layer has an advantage that can cope with severe channel condition such as narrowband interference and frequency-selective fading due to multipath. In MAC layer, 802.11p will use the enhanced distributed channel access (EDCA) mechanism, which is an enhanced version of the distributed coordination function (DCF) from legacy 802.11 standard. EDCA uses carrier sensing multiple access with collision avoidance (CSMA/CA) protocol, CSMA/CA is used to detect the collision problem when packets transmit via wireless mediums. The brief procedure of CSMA/CA is described as follows. Initially, the node with transmitted data will detect the channel status, if the channel is idle for an arbitrary inter-frame space (AIFS) period, and a node is allowed to transmit only at the beginning of each slot time, in which defines as the time need for a node to detect the transmission of other nodes. This time slot includes the propagation delay, the time needed for a node to change from receiving state to transmitting state, and the time to signal to the MAC

layer the state of the channel. After that, the back-off process will be initialized. In the back-off phase, the back-off counter will choose a value uniformly from 0 to contention window size, and begin to decrement whenever the channel is sensed idle, once the counter reaches 0, it will start to send the data. After all, if no ACK frame is sent back from the receiver to the sender, it means the data is lost or packet collision occurs, then the sender should retransmit the data; otherwise if the ACK frame is received by the sender, it means the data is transmitted successfully to the receiver. The more specific procedure of CSMA/CA protocol will be discussed in the next chapter.

We have to notice that in our implemented ETC system, since the range of transaction area is short and the transmitted range of vehicle (300m~1000m in DSRC spec) is large enough to cover with each other, it’s not necessary for us to consider the hidden terminal problem, which implies that the Request to Send/Clear to Send (RTS/CTS) access mode will not be used in our ETC system scenario.

Since 802.11p extends the 802.11e EDCA QoS mechanism, different service classes are obtained by prioritizing the data traffic according to the corresponding access category (AC). There are four queues in MAC layer which implies that each node maintains four queues as illustrated in the Figure 2-4, with parameters that will be used in the queue are such as 𝐶𝐶𝐶𝐶𝑚𝑚𝑚𝑚𝑚𝑚, 𝐶𝐶𝐶𝐶𝑚𝑚𝑚𝑚𝑚𝑚, AIFSN and maximum transmit opportunity (TXOP).

Figure 2-4. The contention between data packets and four queues in MAC layer

The transmitted data with higher priority such as real-time services, will have shorter waiting time in the queue since it has lower contention window size and shorter AIFS, besides, a station can send as many frames as possible in the defined TXOP period; in contrast, the transmitted data with lower priority will have longer waiting time, besides if the corresponding TXOP is 0, it’s limited to transmit a single MAC service data unit (MSDU). Table 2-2 shows the default EDCA parameter for each AC.

Table 2-2. Default EDCA Parameters for each AC

ETC is an automatic toll charging method, which is usually used on the high-way, or on the chargeable tunnels and bridges, also is common for being implemented in the downtown since it can efficiently alleviate the traffic congestion problem in the metropolis area. The fundamental of ETC system is that, whenever a vehicle passes through the toll station, it can use the OBU install on the vehicle to communicate with RSU by the DSRC wireless technology. After all, it will automatically deduct the fee from the user account.

The advantages of implementing ETC system are that, vehicles which install OBU can complete transaction process in a short time without any necessary to prepare for cash or freeway toll tickets; also, it has some features such as energy saving, time saving and alleviation of traffic congestion originally resulted from manual toll collection.

Besides, it’s allowable for ETC system to charge different values of fee in accordance with degrees of traffic congestion. Whenever the congestion is severe, the value of fee should be increased to disperse the traffic flow.

In recent years, many studies of ETC system have been proposed. In [12], the

author mainly reviewed the research and development work of the ETC system, including the composition and principle of ETC system, the design of vehicle’s automatic identify system and several major vehicles’ identification technologies. In [13], the author proposes a MLFF ETC system which has a 6.2m height overhead gantry covering the width of the three lanes motorway, also this system has three antennas based on 5.8 GHz microwave link. In [14], the author proposes a very simple method for enhancing the performance of infrared ETC systems, using two typical low-cost commercial LEDs with different half-intensity angles. In [15], the author proposes a millimeter-wave MLFF ETC system that can simultaneously perform multi-target tracking using the pulse-Doppler radar technique and multi-data communication using the concept of frequency multiplexing in communication systems. In [16], the author proposes an effective method for implementing the enforcement that will be used in the MLFF ETC system.

The general architecture of the ETC system [17] contains two components: (a) Automatic Vehicle Identification and Transaction Processing (b) Violation Enforcement module, which is shown in the figure 2-5. The component of automatic vehicular identification and transaction processing involves RSU and OBUs, when vehicle enters the transmission range of RSU, the RSU will subjectively communicate with the OBU for exchanging information such as authentication and type of vehicles, then RSU will identify the vehicle type and eliminate the fee from its account. Finally, RSU will send the user information to back-end server for further handling of user account.

Figure 2-5 The ETC system architecture

However, if the vehicle does not install the OBU or the fund in the user account is not adequate, the module of violation enforcement will be initialized. The module of violation enforcement involves the digital camera and automatic license plate recognition (ALPR) technology, and the whole operation process is as follows. At beginning, the vehicle image will be derived by the digital camera, then ALPR will recognize the plate number of vehicle. Finally, the plate number information will send back to the back-end server for further pursue of bills.

Our proposed ETC system model is depicted in Figure 2-6. In our system, we divide the transaction area into reservation zone (R-zone) and transaction zone (T-zone), with the central controller contains with the channel allocator and the polling scheduler.

Figure 2-6. The architecture of our developed ETC system

The R-zone is contention-based with a Reservation-RSU (R-RSU) located at the end of area, the purpose for the R-zone is to let the vehicle get the reservation. Based on the channel access scheme of CSMA/CA defined in MAC protocol, once more than one vehicle intends to contend the channel access simultaneously, the reservation might fail resulted from the packet collision. Notice that if a vehicle does not get the channel access before leaving the R-zone, it has a high probability for the transaction process to be uncompleted, hence it’s a critical issue for us to derive the channel reservation ratio of a vehicle. We will discuss about this issue in the next chapter.

The reservation process for a vehicle in R-zone will be described as follows. In CCHI, the R-RSU will periodically broadcast R-beacon frames for each 100ms on CCH, the information contained in the beacon frame are such as timestamp, beacon interval and capability information. When vehicles without channel-reserving receive the beacon message, it will send R-Request (R-REQ) frame to R-RSU on CCH, again we have to notice that there may have several vehicle attempting to send R-REQ packets to R-RSU at the same time, which will result in failure of contention due to packet collision as just mentioned. After R-RSU receives R-REQ frame, the channel allocator

will choose a SCH whose polling queue has the smallest number of un-polled vehicles, the reason for doing so is because of load balancing considerations. When the allocation is done, R-RSU will send back an R-Request (R-REQ) frame, as the vehicle receives the R-REQ message will it tune the SCH to corresponding SCH number and again sends an ACK frame back to R-RSU. At final, R-RSU will add this vehicle to the corresponding polling queue of T-RSU to complete the channel reservation. The packet flow of channel reservation is shown in figure 2-7.

Figure 2-7 Packet flow of channel reservation process

The T-zone is contention-free based with five T-RSUs installed at the end of the area, within one of T-RSUs tuned on CCH, while others poll on SCHs. For a vehicle in T-zone, each T-RSU except the one tuned on CCH, will poll this vehicle on SCH for transaction process, which is based on the operating mechanism of point coordination function (PCF) defined in MAC protocol, that is, these T-RSUs will be the point

coordinators (PC) for vehicles. In addition, the T-RSU which tuned on CCH is used to serve vehicles which are failed to reserve the channel before leaving the R-zone.

The transaction process for a vehicle in T-zone will be described as the follows.

The T-RSU will poll the reserved vehicle as this vehicle is in the T-zone, after vehicles receive the polling frame, they will send back an ACK frame to the corresponding T-RSU based on their own SCH number on SCH. The transaction process is completed when T-RSU receives the ACK packet polled from the vehicle. Notice that the T-RSU will prior to poll the vehicle which has minimum remaining time on T-zone, using the method of earliest deadline first (EDF).

If vehicles cannot reserve channel successfully before leaving the R-zone, the T-RSU which tuned on CCH will use contention-based access scheme to help this vehicle to reserve channel, and use three-way handshake method to complete the whole transaction process. Finally, the flow chart of channel reservation and transaction process is shown in Figure 2-8.

Figure 2-8 Flow chart of channel reservation and transaction process

Chapter 3 Mathematical analysis

In this chapter, we will analyze the successful reservation probability of the R-zone with different parameters, such as the length of reservation R-zone, vehicle arrival rate and average speed of the vehicle. Under our analysis method, the explicit reservation probability with varied length from entrance of R-zone can be estimated.

The remained sections in this chapter will show the top-down approach of the analyzed process.

3.1 Overview of the analysis

Our main objective is to derive the successful reservation probability for a vehicle with different entrance distance. Here, we donate 𝑓𝑓(𝑑𝑑) as the successful reservation probability of a vehicle with distance d from entrance, where d is a real value which is between 0 to the length of reservation zone, and normally uses meters as its unit. As a straightforward thought, a vehicle will get the channel access in the first contention, or in the second contention, even worse in the afterward contentions, hence 𝑓𝑓(𝑑𝑑) can be expressed as following:

The parameters used in this function, including 𝑔𝑔(𝑖𝑖) and 𝑟𝑟(𝑑𝑑), are represented (1)

as the successful probability for 𝑖𝑖-th channel access and the maximum times of contentions for a vehicle (or the maximum transmission times of the R-REQ sent by a vehicle) with distance 𝑑𝑑, respectively. Figure 3-1 illustrates the schematic graph for vehicle to reserve the channel.

In the following sections 3.2-3.4, we will analyze the calculation methods of 𝑟𝑟(𝑑𝑑) and 𝑔𝑔(𝑖𝑖) step by step.

Figure 3-1 Overview of the reservation process for a vehicle

3.2 The analytical steps for transmission times

In the analysis of 𝑟𝑟(𝑑𝑑), we will introduce two parameters ℎ𝑐𝑐(𝑖𝑖) and ℎ𝑠𝑠(𝑖𝑖) at first. The former one ℎ_𝑐𝑐(𝑖𝑖) is vehicle displacement for the 𝑖𝑖-th contention that has been in failure, and the latter one ℎ𝑠𝑠(𝑖𝑖) is the vehicle displacement for the 𝑖𝑖-th contention that has been in success. If a vehicle intends to reserve channel successfully by distance 𝑑𝑑, it must at least get the channel successfully in the last contention, in which means the last time for this vehicle to contend the channel access within distance 𝑑𝑑. This also indicates that, this vehicle losses all contentions before the last successful contention. Through the schematic diagram shown in the figure 3-2, we will clearly understand the architecture of the analysis.

Figure 3-2. The schematic diagram of the system model

Hence, 𝑟𝑟(𝑑𝑑) can be computed as follows. If there exists a real positive integer α that satisfies ℎ_𝑐𝑐(1) + ℎ_𝑐𝑐(2) + ⋯ +ℎ_𝑐𝑐(α − 1) + ℎ_𝑠𝑠(α) < 𝑑𝑑 and ℎ_𝑐𝑐(1) + ℎ_𝑐𝑐(2) +

⋯ + ℎ𝑐𝑐(α) + ℎ𝑠𝑠(α + 1) > 𝑑𝑑, then the maximum contention times 𝑟𝑟(𝑑𝑑) is equal to α.

Formally, it can be expressed as:

∃𝛼𝛼 ∈ 𝑁𝑁 𝑠𝑠. 𝑡𝑡. � ℎ𝑐𝑐(𝑖𝑖) + ℎ𝑠𝑠(𝛼𝛼) Assume that the vehicle speed is a constant, so in the rest of this section, there are only two parameters that we have to estimate, which are failed transmission time 𝑇𝑇_𝑐𝑐(𝑖𝑖) due to packet collision and successful transmission time 𝑇𝑇𝑠𝑠(𝑖𝑖), respectively.

In the IEEE 802.11 series standards, the channel access delay for a node to transmit the data is based on CSMA/CA mechanism defined in MAC protocol and is composed of the following three components: the medium access time after a busy medium, back-off delay before channel access and data transmission delay.

1. The medium access time after a busy medium: When a node attempts to access the channel, it has to detect the current channel status through executing the function of clear channel assessment (CCA) to detect whether the channel is in idle or busy condition. If the channel is in idle status, it can immediately access the channel after waiting for a period; on the other hand, if the channel is in busy condition, which means there is one of other nodes using the channel, it has to wait for the channel to become idle, then also have to wait for a period to begin the back-off procedure. Since the contention-based mechanism in the MAC layer of WAVE/DSRC follows 802.11e EDCA, the period can be expressed as:

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴[𝐴𝐴𝐶𝐶] = 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 + 𝑎𝑎𝐴𝐴𝑎𝑎𝑎𝑎𝑡𝑡𝑇𝑇𝑖𝑖𝑎𝑎𝑎𝑎 × 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑁𝑁[𝐴𝐴𝐶𝐶] (3)

The AIFS means “arbitrary inter-frame space”, this parameter is similar to the DIFS of legacy DCF of IEEE 802.11 standard, but due to several priority levels in the EDCA, it has different values of AIFS according to the selected access category (AC).

2. Back-off delay before channel access: When a node begins a back-off procedure, it will uniformly select a random value between 0 to CW-1, where CW is the current contention window (CW) size. After that, it begins the back-off process within the value decremented, once the back-off counter reaches 0, the node will start to transmit the data. We have to notice that the CW size is not a fixed constant, as the node begins to transmit data without any retransmission, the CW size will be set as the minimum of contention window (generally donated as 𝐶𝐶𝐶𝐶_{𝑚𝑚𝑚𝑚𝑚𝑚}), as the retransmission occurs, the CW size will be doubled until reaching the maximum of contention window (generally donated as 𝐶𝐶𝐶𝐶_{𝑚𝑚𝑚𝑚𝑚𝑚} ), when the CW size

reaches 𝐶𝐶𝐶𝐶𝑚𝑚𝑚𝑚𝑚𝑚, it will stop to increment. In the regulation of 802.11 series, every packet can be retransmitted by a node with limited times, known as Station Short Retry Limit (SSRL), once the packet has been retransmitted with limited times, the

packet should be dropped, within the CW size reset to 𝐶𝐶𝐶𝐶_{𝑚𝑚𝑚𝑚𝑚𝑚}, then the packet is allowed to rejoin the contention of channel access. The whole process is known as binary exponential back-off mechanism.

3. Data transmission delay: In the initiation of this phase, a vehicle will send R-REQ message to the R-RSU, after that, there will have two possible cases, which are failed transmission and successful transmission, respectively. For the former case, if the vehicle does not receive the R-RES message sent from the R-RSU for a DIFS period, it means that the collision occurs during the data transmission. In this kind of situation, the R-REQ message sent from vehicles should be retransmitted; for the latter case, if R-REQ message has been successfully received by the R-RSU, the R-RSU will send R-RES message to the vehicle after a SIFS period. Again the vehicle will send an ACK message back to the R-RSU, so that R-RSU can confirm the connectivity between them is in a good status, and the channel allocator will add this vehicle to the corresponding polling queue of SCH. After all, the channel reservation process is completely done.

Figure 3-3 shows the IEEE 802.11e EDCA basic access mechanism, the steps for channel access are the same as the previously mentioned three components.

Figure 3-3. The IEEE 802.11e EDCA basic access mechanism propagation delay, and 𝑝𝑝𝑓𝑓𝑟𝑟(𝑖𝑖) stands for the freezing probability for the back-off counter of 𝑖𝑖 -th contention. In addition, 𝐶𝐶𝐶𝐶(𝑖𝑖) and 𝑑𝑑_{𝑠𝑠𝑚𝑚𝑠𝑠𝑠𝑠} will be explained as follows:

 𝐶𝐶𝐶𝐶(𝑖𝑖): The contention window size might have a variation as the retransmission occurs, in the following equation, the parameter of m represents the number of times which CW size increments from 𝐶𝐶𝐶𝐶_{𝑚𝑚𝑚𝑚𝑚𝑚} to 𝐶𝐶𝐶𝐶_{𝑚𝑚𝑚𝑚𝑚𝑚}, and M represents the maximum retransmission limit. Hence we express 𝐶𝐶𝐶𝐶(𝑖𝑖) as:

𝐶𝐶𝐶𝐶(𝑖𝑖) = � 2^𝑚𝑚∙ (𝐶𝐶𝐶𝐶𝑚𝑚𝑚𝑚𝑚𝑚+ 1) 𝑓𝑓𝑎𝑎𝑟𝑟 0 ≤ 𝑖𝑖 ≤ 𝑎𝑎

2^𝑚𝑚∙ (𝐶𝐶𝐶𝐶_{𝑚𝑚𝑚𝑚𝑚𝑚}+ 1)𝑓𝑓𝑎𝑎𝑟𝑟 𝑎𝑎 < 𝑖𝑖 ≤ 𝑀𝑀, m = log2𝐶𝐶𝐶𝐶_{𝑚𝑚𝑚𝑚𝑚𝑚}+1

𝐶𝐶𝐶𝐶_{𝑚𝑚𝑚𝑚𝑚𝑚}+1, 𝑀𝑀: 𝐴𝐴𝐴𝐴𝑅𝑅𝑆𝑆 (6)

 𝑑𝑑𝑠𝑠𝑚𝑚𝑠𝑠𝑠𝑠: Every time the back-off counter decrements, the node will continuously detect the channel status, if the channel is staying in idle state, then the back-off counter will keep decrementing; in contrast, if the channel is changed into busy state, the back-off counter will cease to decrement, waiting for the completion of data transmission from other nodes, then the back-off counter will restore to the decremented process. Here, the probability that the back-off counter will cease to decrement due to channel states changes into busy is called “frozen probability”, we donate it as a function 𝑝𝑝𝑓𝑓𝑟𝑟(𝑖𝑖), and the delay of data transmission by other nodes is donated as a function𝑑𝑑_{𝑠𝑠𝑖𝑖𝑠𝑠𝑎𝑎}. Thus, 𝑑𝑑𝑠𝑠𝑚𝑚𝑠𝑠𝑠𝑠 can be expressed as:

𝑑𝑑𝑠𝑠𝑚𝑚𝑠𝑠𝑠𝑠 =3𝑅𝑅

𝐷𝐷 + 2 ∙ 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 + 2 ∙ 𝛿𝛿 (7)

3.3 The Markov chain model

Before conducting the successful probability of channel contention, we utilize the Markov chain model to describe the process of contending channel at first.

We consider a scenario that there are always n stations contending for the channel access, which means once a frame is successfully transmitted by a station, there will have another frame which waits for being transmitted, immediately. We give that 𝑏𝑏(𝑡𝑡) be the back-off window size at random time 𝑡𝑡 in the process of channel contention, and 𝑠𝑠(𝑡𝑡) be the back-off stage (0,1,…,M) at random time 𝑡𝑡, where 𝑀𝑀 stands for station short retry limit (SSRL).

When the back-off counter reaches 0, a station will send out the frame. However, if there are other stations transmitting the data at the same time, then the packet collision will occur, and we donate 𝑝𝑝 as the collision probability. If we

donate [s(t), b(t)] as the status of channel contention, then the process of channel contention will be a discrete Markov-chain which is shown in the Figure 3-4, with the one-step transition probabilities as follows:

⎩⎪

⎨

⎪⎧𝑃𝑃{𝑖𝑖, 𝑘𝑘|𝑖𝑖, 𝑘𝑘 + 1} = 1 − 𝑝𝑝𝑓𝑓𝑟𝑟 𝑘𝑘 ∈ (0, 𝑐𝑐𝑐𝑐(𝑖𝑖) − 2); 𝑖𝑖 ∈ (0, 𝑀𝑀) 𝑃𝑃{0, 𝑘𝑘|𝑖𝑖, 0} = 1

𝑐𝑐𝑐𝑐(0) 𝑘𝑘 ∈(0, 𝑐𝑐𝑐𝑐(0) − 1); 𝑖𝑖 ∈ (0, 𝑀𝑀) 𝑃𝑃{𝑖𝑖, 𝑘𝑘|𝑖𝑖 − 1,0} = 𝑝𝑝

𝑐𝑐𝑐𝑐(𝑖𝑖) 𝑘𝑘 ∈(0, 𝑐𝑐𝑐𝑐(𝑖𝑖) − 1), 𝑖𝑖 ∈ (1, 𝑀𝑀)

(8)

Among which: (1) The decrement of the back-off time counter; (2) After a packet is transmitted successfully at stage i, a new packet begins with back-off stage 0; 3) After a packet is transmitted unsuccessfully at back-off stage 𝑖𝑖, the back-off interval is uniformly chosen in the range (0, cw(i + 1)).

在文檔中 WAVE/DSRC電子收費系統在預約區之分析與驗證 (頁 17-0)