2017 9th IEEE International Conference on Communication Software and Networks
An Adaptive TDMA Scheduling Strategy Based on Beacon Messages for Vehicular Ad-Hoc Network
Chaodong Wu, Bin-Jie Hu, Zong-Heng Wei, Lizhao Lin School of Electronic and Information Engineering South China University of Technology Guangzhou, China e-mail:
[email protected],
[email protected], {wei.zongheng, eelzlin}@mail.scut.edu.cn Abstract—Aiming at dynamic changes of network topology, high real-time security services and high network load in VANETs, this paper proposes an adaptive TDMA slot allocation strategy based on beacon messages (BMA strategy). BMA strategy makes full use of neighbor nodes’ beacon messages, including vehicular direction, position, velocity, etc. Meanwhile, each vehicle establishes neighbor table containing beacon messages. By using time slot acquisition algorithm, neighbor table update algorithm and BMA slot allocation strategy, the related model is established. BMA strategy can predict time slot allocation of next frame according to node density ratio. Furthermore, vehicles with similar velocity could acquire near slots, this mechanism reduces the rate of merge collisions. Comparing with ADHOC and VeMAC protocols, BMA strategy has higher time slot competition success rate, less node collision and better scalability. Keywords-VANET; MAC protocol; TDMA; beacon messages; time slot allocation
I.
INTRODUCTION
In recent years, car ownership has grown rapidly. Intelligent Transportation Systems (ITS) have gradually developed, and become a research hotspot in the academia and engineering. Vehicular Ad-hoc Network (VANET) is regarded as the foundation and core technology of developing ITS and improving traffic safety. VANET is a new wireless ad-hoc network consisting of moving vehicles and road side units (RSUs), which has the capabilities of perception, calculation, storage and wireless communication. Comparing with traditional mobile ad-hoc network, VANET has the characteristics of high dynamic change in network topology, real-time traffic safety and high network load. Research on how to access wireless channel and how to use channel resources effectively in MAC protocol for VANET plays a decisive role on many indexes (i.e. delay, throughput and channel utilization). The ideal MAC protocol can effectively solve the problems such as hidden terminal, exposed terminal, access collision, merge collision, etc. In accordance with the different ways of wireless channel access coordination, the current MAC protocol in the network is mainly divided into two types based on competition and non-competition. CSMA/CA and TDMA are two commonly used technologies. Due to rapid movement of vehicles, the distribution of traffic flow on road
978-1-5090-3822-0/17/$31.00 ©2017 IEEE
255
is uneven and changing dynamically. MAC protocol based on CSMA/CA mechanism has message collision when the number of vehicles increases, while MAC protocol based on TDMA mechanism has low channel utilization when the traffic flow is sparse. Time Division Multiple Address (TDMA) mechanism can dispatch nodes to transmit data at different time slots, then provide data packets transmission without collisions. Therefore, it is a breakthrough to solve the problem by designing a TDMA-based MAC protocol that is efficient, stable and adaptive. Standard [1, 2] analyzes the performance of 802.11p in detail, it comes to a conclusion by a large number of simulation experiments that 802.11p protocol cannot guarantee real-time data transmission and bounded delay under the condition of large data flow. ADHOC protocol [35] is based on RR-ALOHA channel access mechanism, and uses TDMA method to divide wireless channel. This protocol is suitable for VANETs, yet the drawback is that access collision occurs when nodes move faster and node density becomes larger. VeMAC protocol [6-8] divides time slot into three slot sets, and nodes select competition slots randomly in the corresponding time slot sets, which effectively solves the problem of access collision and merge collision. However, when node density in the network is larger and the nodes move faster, it will lead to the increase of access delay and the decline of channel utilization. A special VANETs multi-channel MAC protocol [9] (DMMAC) uses an adaptive broadcast mechanism to reduce message collision and transmission delay of security information. DMMAC enhances the probability of secure packet delivery by varying CCH interval length. However, dynamic adjustment of CCH interval length is not taken into account. A variable CCH interval (VCI) protocol and multichannel MAC support multi priority access mechanism was proposed in the literature [10], which calculates the time of reserving information above CCH in saturated condition, so as to achieve the optimal CCH interval and adjust period of CCH/SCH under different traffic flow density. Whereas unsaturated state is not considered. When dynamic topology changes in vehicular network, the performance of traditional MAC protocol based on fixed time slot will drop, which is not conducive to transmit security services efficiently. This paper proposes an adaptive TDMA slot allocation strategy based on beacon messages (BMA strategy) to solve the problem. In Section II we
elaborate system model and design structure of proposed strategy. In Section III we introduce the implementation of scheduling strategy by using three algorithms. Section IV gives the simulation results and analysis while Section V comes to a conclusion. II.
frame, we give priority to slot allocation for low vehicle density area. The TDMA frame structure is shown in Fig. 2. One time frame is divided into two parts, namely low density areas L, high density areas H, and two part of the slot is not fixed, but according to dynamic topology of real scene. Each vehicle establishes a neighbor table, which will update realtime information according to the density ratio of vehicles in left lane direction and right lane direction. By using adaptive algorithm, the strategy calculates time slot allocation proportion of high density area and low density area for next frame, which leads to dynamic time slot allocation and satisfies complex network environment. In order to reduce merge collisions effectively, we divide slots according to vehicular velocity in each density zone. Vehicles can broadcast periodically within a THS, different vehicles enjoy the same time frame. While in different THSs, change of vehicular velocity causes change of node position, vehicles from different THSs may change to the same THS, e.g., two vehicles running opposite or driving in the same direction but the velocity of rear vehicle is greater than the front one, then merge collision occurs. The slots of same density areas is divided into two slot sets. We take the median of node velocity as the critical value, dividing into low speed zone and high speed zone, denoted by v and v’. Vehicles with similar velocities acquire the near slot. Thus, nodes with large difference in speed can acquire suitable slot, which reduces the rate of merge collisions.
SYSTEM MODEL AND STRUCTURE DESIGN
A. System Model In real life, most of the vehicular moving scenes are twoway traffic road scenes. According to vehicular direction, vehicles are divided into two categories, as shown in Fig. 1. According to geographical position, vehicles traveling east or easterly direction are defined as the right category, the opposite are defined as the left category. Each vehicle is equipped with a global positioning system (GPS) receiver and can accurately determine its position and moving direction. Synchronization among vehicles is performed using the 1PPS signal provided by any GPS receiver. In case of a temporary loss of GPS signal, the synchronization among different nodes can still be maintained within a certain accuracy for a time duration, which depends on the stability of the GPS receiver’s local oscillator.
Figure 1. Vehicular moving scene.
In the system model, total vehicles are divided into two sets within a two-hop set (THS), including low density areas and high density areas, respectively denoted by S L ,SH . A THS is a set of nodes in which each node can reach any other node in two hops at most. Node i is denoted by ni , then two sets are expressed as: SL {ni | i 1, 2,..., N L } (1)
SH
{ni | i 1, 2,..., N H }
Figure 2. TDMA frame structure.
The problem of access collision and merge collision will occur when nodes acquire slot. Access collision occur when two or more nodes compete for the same channel. If access collision occurs, nodes will choose to wait until the next competition period arrival, so that access delay will increase, as shown in Fig. 3.
(2)
B. Frame Structure Design Considering the uneven distribution of traffic flow on real road, e.g., vehicular density in left lane is greater than that in right lane, at this point, more slots should be allocated to high density areas, so that the majority of nodes can quickly acquire slots. In this way, the channel utilization is improved and the resource allocation is more reasonable. Considering the motion state of vehicles in low density regions is more complicated, e.g., vehicles accelerate or overtaking, and transmission of vehicular security service needs to be safeguarded, low density areas should be the priority scheduling at this time. In the structure design of one
Figure 3. Access collision caused by node competition.
Nodes beyond one THS can use the same time slot without collision due to spatial multiplexing. With nodes moving, merge collision will occur when two nodes move to
256
the same THS, which causes that node communication failed and nodes compete for channel again, shown in Fig. 4.
density area and high density area is
TLi THi . The following rules are considered in the system model to make the model more accurate: Rule 1: The vehicles are located within a THS, which can acquire information by broadcasting neighbor nodes’ beacon messages. Rule 2: If a slot is not occupied by nodes, it is considered to be an available slot, and nodes can obtain it at any time. Rule 3: If a slot is acquired by two or more nodes, it’s considered to be a collision, meaning these nodes obtains slot unsuccessfully, so it’s necessary to obtain the slot again. Ti
Figure 4. Merge collision caused by node mobility.
C. Neighbor Table Definition In vehicular environment, each vehicle maintains its own vehicular beacon messages, including direction, velocity, position, etc. And neighbor nodes obtain beacon messages by other nodes broadcasting within a THS. Each node establishes and updates neighbor table according to network load. Neighbor table can acquire nodes beacon messages and calculate network load, the ratio of high density area and low density area, slot allocation ratio, etc. Then the strategy completes the process of a series of slot reservation and slot access. Thus, the frame structure adjust dynamically, which adapts to the real network environment and has good scalability. The neighbor table of node i is marked as Wi , including a list of neighbor nodes information, e.g., Wi1 ,Wi 2 , etc. Take the low density area as an example, represented as follows:
III.
{Wij | j 1, 2,..., N L }
IMPLEMENTATION OF SCHEDULING STRATEGY
A. Time Slot Acquisition Considering the process of obtaining time slots in a THS, assuming there are K nodes and N slots in a frame, we can establish a related model to get the probability that a node acquire a time slot within n frames and the average number of nodes acquiring a slot. Assuming the probability that each node preparing for slot competition is P, and total number of successful nodes after n frames is M, then M is a discrete time Markov chain [8], as show in Fig. 5, the transfer probability is calculated as follows:
Pi , j
Wi
TLi , THi , so
(3)
1, i j k ° ° S ( j i, K i, N i ) , 0 d i d K 1, i d j d k ® ( N i ) K i ° ° ¯0, others
(5)
Among them, each Wi contains five types of information, they are respectively direction type, density type, velocity, velocity type and amount, which is denoted by:
Wij
{DirectionType, DensityType,Velocity,
VelocityType, Amount}, j 1, 2,..., N L
(4) Figure 5. Markov state transition chain.
DiretionType is a Boolean variable, when it takes 0 it represents vehicles in left lane while 1 represents vehicles in right lane. DensityType is a Boolean variable, when it takes 0 it represents vehicles in low density area while 1 represents vehicles in high density area. Velocity is a floating point variable, representing the true velocity, denoted by vi . VelocityType is a Boolean variable, when it takes 0 it represents the low-speed vehicles while 1 represents the high-speed vehicles. Amount is an integer variable, representing the total amount of vehicles in high or low density area, denoted by N L , N H . In addition, the frame
Among them, S (m, k , n) is the total number of m nodes acquiring the time slots successfully, m is the number of successful nodes, k is the total number of competing nodes, n is the total number of valid slots. According to regulation of permutation and combination, the S (m, k , n) function is calculated as follows: k m m m ª º k m °Ck Cn m ! «(n m) ¦ S (i, k m, n m) » ,0 d m k i 1 ¬ ¼ ° ° m S (m, k , n) ®Cn m !, m k (6) °0, m ! k ° °¯
length corresponding to the node i is denoted as Ti , namely the total number of slots, at this time slot number for low
257
The number of nodes competing successfully is denoted by X n , then P( X n
i)
i
¦P
ni
First initialize the information, the DirectionType of each node is determined according to vehicular direction. Nodes update personal DirectionType information, e.g. 0 or 1. The Velocity of each node is updated according to its own velocity, then nodes broadcast messages, so that the neighbor table of each node within a THS is initially established, containing beacon messages of other neighbor nodes. The DensityType of node needs to be gained according to Amount in the neighbor table, nodes are divided into low density areas and high density areas within a THS, when Amount of one density region is greater than the other region, namely high density area, this moment the DensityType is 1, while the opposite is 0. The VelocityType of node is calculated by the node velocity, we make nodes velocity as a collection, sort the velocity data, and find the median vm , the
, all nodes access to time slot
n 0
when i k . The average number of nodes acquiring time slots within n slots is calculated as follows:
Pn
k
P( X ¦ i P(
n
i)
(7)
i 0
The probability that each node successfully obtains the slot within n slots is calculated as follows:
Ps
Pn K
(8)
VelocityType is 1 when velocity is higher than vm , while the opposite is 0. The Amount of the neighbor table changes dynamically according to vehicular network, when the DirectionType of nodes changes, e.g., change from 1 to 0 or from 0 to 1, namely vehicles turn around, at this time the Amount of current density region minuses 1, and another one pluses 1, adjusting DensityType at the same time. Algorithm processes are written as follows:
Simulate the process of node obtaining time slot, each node obtains one available slot randomly when competition begins. The current time slot state becomes 0 and the collision node state changes to 0 if collision occurs, and nodes compete time slot after a frame. Time slot acquisition algorithm is presented as follows: Algorithm 1. Time slot acquisition Initialize(); Nstate=zeros(1,N); Kstate=zeros(1,K); while not all nodes acquire time slot for i= 1 to K if Kstate(i)=0 Kstate(i)=rand(available slots); end if end for for j= 1 to N if slot j is acquired Nstate(j)= Nstate(j)+1; end if end for for i= 1 to K if node i collides Kstate(i)= 0; end if end for end while
Algorithm 2. Neighbor table update Initialize()˗ if Amount>Amount’ DensityType =1; else DensityType =0; end if vm=median(v1,v2,…); for i=1 to Amount if v(i)>=vm VelocityType=1; else VelocityType=0; end if if DirectionType change Amount= Amount-1; Amount’= Amount’+1; end if end for C. BMA Slot Allocation Strategy We get the neighbor tables of all nodes in a THS, and update neighbor table in real time. It can predict time slot allocation of next frame by using neighbor table information, which will satisfy vehicular network topology changing. Main steps of BMA slot allocation strategy are: 1) Count neighbor table information of each vehicle, and obtain the Amount value of the neighbor table, vehicles are divided into low density areas and high density areas within the THS, the corresponding Amount value is N L , N H .
B. Neighbor Table Update Considering the process of nodes establishing and updating neighbor tables, we know from previous definition of neighbor table: Wij
{DirectionType, DensityType,Velocity,
2) Calculate the ratio of N L and N H , that is H
VelocityType, Amount}, j 1, 2,..., N L
NL , NH
dynamically adjust the time slot number of high density area
258
within n frames in ADHOC, VeMAC protocols and BMA strategy. The red lines represent the case (K=16, N=24) of three strategies while the blue lines represent the case (K=12, N=14). It is observed that the performance of proposed strategy is better than ADHOC and VeMAC. Especially, while the number of nodes and slots is close (K=12, N=14), the average successful number is larger obviously when using proposed strategy because of less collisions. Fig. 8 shows the probability that all nodes acquire a time slot within n frames in ADHOC, VeMAC and BMA strategy. We can learn that with a probability around 0.9, all nodes acquire a time slot within three frames for the case (K=12, N=14) in BMA strategy, while four frames in VeMAC and five frames in ADHOC. BMA strategy divides a frame into two part and divides each part into two small slots according to node velocity, so that it can reduce access delay.
and low density area, make it satisfying: TL | N L . And the TH
TLi THi , then
frame length meet the condition: Ti TLi
ª N L º , so T Hi «Ti » N L NH ¼ ¬
NH
Ti TLi .
3) It shall ensure that both sides of time slots have certain resource when N L and N H differ greatly. The slot ratio is specified as 10% if H 10% , that is TL / TH 10% , so as to reserve a certain channel resources to solve the problem of vehicular network accident. BMA slot allocation strategy is presented in Algorithm 3. Algorithm 3. BMA slot allocation strategy Initialize()˗ if NL/NH-TL/TH
