Performance evaluation of contact-time based and adaptive-timer based message suppression methods for inter-vehicle communication in vehicular DTN

Abstract

In the next generation wireless networks, the number of connected terminals to the network, communication protocols, and the channels available will be increased, thus network slicing will become more important. Also, vehicles, buses, trains and motorcycles are considered communication terminals. These communication terminals should have independent network management considering their movement such as joining and leaving the networks. Therefore, Delay-Disruption-Disconnection Tolerant Networking (DTN) has been attracting attention for their potential support of inter-vehicle communication. In this paper are presented the Contact-Time (CT) based and Adaptive-Timer (AT) based Message Suppression (MS) methods for Vehicular DTN. For the CT-based MS method are used three DTN protocols for Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communication. For AT-based MS are used conventional Epidemic and two proposed Epidemic-based methods for V2V communication. We compare MS method, Message Suppression Controller (MSC) and MSC with Adaptive Threshold (MSC-ATh). The simulation results show that MSC-ATh performs better than other approaches. The storage consumption is improved when the number of vehicles increases and there is no reduction in PDR even if the message suppression is enabled. For Epidemic, when the number of Road-Side Units (RSUs) is 16, the results of PDR are the best compared with other DTN protocols. The MSC-ATh method is about 22% better than Epidemic for storage consumption. Also, the delay performance of MSC-ATh is improved by increasing the Suppression Coefficients (SCs) and number of vehicles.

Keywords

DTN Adaptive Timer VANET Message Suppression

1. Introduction

In order to realize a resilient and vigorous society, the information and communication network infrastructure based on Beyond 5G should be inclusive, sustainable and highly dependable [3,24]. Inclusive life requires a network system that supports the infrastructure of a society. The network should be designed to have low loss, convenience and sustainable growth. Moreover, the future networks should support human-centered society where security and safety are maintained and trust in a relationship is strong even when unexpected situations such as COVID-19 pandemic occur.

In the future wireless networks, vehicles, buses, trains and motorcycles are considered communication terminals. With the rapid improvement in the communication area covered by a single terminal, the core network at the edge will become more important to control. Also, the number of connected terminals to the network, communication protocols, and the channels available will be increased, thus network slicing becomes more important. These communication terminals should have independent network management considering their movement such as joining and leaving the networks. Therefore, Delay-Disruption-Disconnection Tolerant Networking (DTN) has been attracting attention for their potential support of inter-vehicle communication [6–8,12,19,30]. Many researchers and developers have proposed several algorithms for message ferrying [1,5,9,14–17,20,23,26,31,32,34,35]. Due to the overhead of the conventional DTN approach, the consumption of network resources and storage become critical issues. Therefore, recovery approaches have been recommended as possible solutions to solve these problems [10,13,27,28,33].

In this paper, we present the Contact-Time (CT) based and Adaptive-Timer (AT) based Message Suppression (MS) methods for Vehicular DTN. For the CT-based MS method, we use three DTN protocols for Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communications. For the proposed AT-based MS, we use conventional Epidemic and two enhanced Epidemic-based methods for V2V communication. The simulation results show that Message Suppression Controller with Adaptive Threshold (MSC-ATh) performs better than other approaches. The storage consumption is improved when the number of vehicles increases and there is no reduction in Packet Delivery Ratio (PDR) even if the message suppression is enabled. For Epidemic, when the number of Road-Side Units (RSUs) is 16, the results of PDR are the best compared with other DTN protocols. The MSC-ATh method is about 22% better than Epidemic for storage consumption. Also, the delay performance of MSC-ATh is improved by increasing the Suppression Coefficients (SCs) and number of vehicles.

The structure of this paper is as follows. In Section 2 is introduced the related work. In Section 3 is explained the simulation modeling. In Section 4 is presented the proposed CT-based MS method and its evaluation. In Section 5 is explained the proposed AT-based MS methods for inter-vehicle communications and their performance evaluation. In Section 6 are given conclusions and future work.

2. Related work

2.1. DTN

The DTN can provide sustainable internet connectivity for space missions [7,19,30]. The space networks are susceptible to link failure, disconnection and significant delay. The intermediate nodes in Vehicular DTN keep bundle messages in their storage before transmitting them to other nodes. The RFC 4838 [4] specifies the network architecture.

Presently, many DTN protocols such as Epidemic [18,29], Direct Delivery, Spray and Wait (SpW) [25] and MaxProp [2] protocols are proposed. The Epidemic protocol is performed by using two control messages to replicate a bundle message. The bundle protocol is specified in RFC 5050 [22]. Each node periodically broadcasts a Summary Vector (SV) in the network. The SV contains a list of stored messages for each node. When the nodes receive the SV, they compare the received SV to their SV. The node sends a REQUEST message if the received SV contains an unknown bundle message. In this method, consumption of network resources and storage state become critical issues, because the nodes replicate bundle messages to adjacent nodes in their communication range. Then, the received bundle messages remain in the storage and the bundle messages are continuously replicated even if the end-point receives the messages. Thus, recovery schemes like timers and anti-packets may delete replicated bundle messages in the network.

Due to the high latency of the sparse network, the anti-packet deletes duplicated messages too late. In the typical anti-packet, the endpoint broadcasts the anti-packet, which comprises a list of received bundle messages. Then, the nodes remove messages and duplicate the anti-packet. In the case of the typical timer, the messages have a lifetime and are removed at regular intervals after their lifetime has elapsed. Therefore, an adaptable lifetime algorithm is needed for utilizing in various environments.

The Direct Delivery does not send a copy of the bundle message to intermediate vehicles. It sends the bundle messages to the end-point directly. The SpW achieves its goal of optimizing network resources by imposing a strict upper limit on the number of copies of each bundle message allowed in the network. The procedure is broken down into two distinct phases: the spray phase and the wait phase. When compared to Epidemic, it is possible to reduce the cost of communicating with others.

The MaxProp is able to solve the problem of storage by prioritizing not only the schedule of packets to be dropped, but also the schedule of packets to be transmitted to other vehicles. The MaxProp will send packets with a high priority when it makes contact with a vehicle that is adjacent to it. These packets will be sorted according to their hop count. When the storage becomes full, the MaxProp will discard the packets with a low priority. These packets are arranged according to the likelihood that they will be delivered.

2.2. Inter-Vehicle Communication based on V2V and V2I

The IEEE 802.11p is an approved amendment to the IEEE 802.11 standard that supports Wireless Access in Vehicular Environments (WAVE). It specifies 802.11 enhancements required to support applications for intelligent transportation systems. The 802.11p standard is based on the IEEE 802.11 architecture, but version “p” is designed for V2V and V2I communication in Vehicular Ad-Hoc Networks (VANETs). This technology uses the 5.9 GHz frequency band in different places such as cars, open spaces and downtown. This standard says that WAVE is the signaling method and interface functions that are controlled by physical layer devices in situations where physical layer properties change quickly and information exchanges are short. This standard gives a set of rules for making sure that wireless devices can talk to each other in environments and times that change quickly.

3. Simulation modeling

We implemented the proposed methods in Scenargie [21] simulator. We carried out the simulations using Scenargie and send bundle messages between start-point and end-points. The number of end-points and support for mobility depend on the scenario and are explained in the following sections.

3.1. Setup for CT-based MS method

In Fig. 1, we show a grid road model with RSUs to evaluate our proposed CT-based MS method. We consider an urban grid scenario with 300 vehicles in the network. Twenty RSUs are built as shown in Fig. 1. When there are four RSUs, they are distributed evenly around the region at the four corners. Simulation parameters are shown in Table 1.

Fig. 1.

Model of grid roads considering 300 vehicles and maximum 20 RSUs.

Table 1

Simulation parameters for CT-based MS method

Parameter	Method/Value
Simulation Time ( $T_{max}$ )	600 seconds
Area Dimensions ( $X \times Y$ )	1500 m $\times 1500$ m
Building: Height	10 meters
Road Width	15 meters
Number of RSUs	0 (N/A), 4, 8, 12, 16, 20
Number of Vehicles	300 vehicles
Mobility Model	GIS-based Random Way-point
Vehicles: Minimum Speed ( $V_{min}$ )	30 km/h
Vehicles: Maximum Speed ( $V_{max}$ )	60 km/h
DTN Protocols	Direct Delivery, SpW, Epidemic
Number of Start-point (fixed position)	1 (Fixed)
Number of End-points (vehicles)	150 vehicles (Random Selection)
Bundle: Message Size	500 bytes
Bundle: Sent Interval	10 seconds
Frequency Bandwidth	5.9 GHz
Physical Model	IEEE 802.11p
Radio Propagation Model	ITU-R P.1411
Antenna Model	Omni-directional
Antenna Height	1.5 meters

3.2. Setup for AT-based MS methods

We consider a scenario within an urban grid with 50, 100 and 150 vehicles/km². Within the structure of the road model, we consider the buildings at each corner (see Fig. 2). In this scenario, RSU is committed in ten percent of the intermediate vehicles. In our simulations, we considered IEEE 802.11p standard, which is intended to provide a set of specifications with the goal of ensuring interoperability between wireless devices that can communicate in environments that are prone to rapid change during specific time periods. Table 2 shows the parameters that are used to evaluate the AT-based MS methods. The proposed AT-based methods do not depend on RSU. The following proposed methods are based on Epidemic protocol and send bundle messages from start-point to end-point for different parameters.

Message Suppression Controller (MSC)

MSC with Adaptive Threshold (MSC-ATh)

The total distance between the start-point and the end-point is about

1, 414

meters. We analyze the performance of the network by considering delay, duplicated bundle messages, transmitted data and the percentage of packets that are delivered. During the simulation, the start-point will send a total of forty bundle messages to the end-point. We consider a Geographic Information System (GIS)-based mobility model based on Random Way-Points (RWP). In this model, the mobility of vehicles is restricted based on the number of congested vehicles and simplified traffic control mechanisms.

Fig. 2.

Road model with fixed start-point and end-point.

Table 2

Simulation parameters for AT-based MS method

Parameter	Method/Value
Simulation Time ( $T_{max}$ )	600 sec
Area Dimensions ( $X \times Y$ )	1000 m × 1000 m
Building: Height	10 meters
Number of Vehicles	50, 100, 150 vehicles/km²
Number of MSCVs	5, 10, 15 (10% of vehicles)
Mobility Model	GIS-based random way-point
Minimum Speed ( $V_{min}$ )	0 km/h
Maximum Speed ( $V_{max}$ )	60 km/h
Routing Method	Epidemic, MSC, MSC-ATh
MSC: MS Threshold	1 (conventional) or Adaptive threshold method
MSC: Refresh Interval	2 seconds
MSC: MST	30, 60, 120 seconds
Application	Bundle Message
Number of Start-point	1 (Fixed)
Number of End-point	1 (Fixed)
Bundle: Start and Stop Time	10–410 seconds
Bundle: Message Send Interval	10 seconds
Bundle: Message Size	500 bytes
PHY model	IEEE 802.11p
Frequency	5.9 GHz
Propagation model	ITU-R P.1411
Antenna model	Omni-directional
Antenna height	1.5 meters

3.3. Mobility model for inter-vehicle communication

Nowadays, the RWP mobility model is utilized in the large majority of simulations. In RWP mobility model, every node is considered mobile and the nodes go from one way-point to another way-point. The following guidelines are followed by every node in the network.

The node determines an initial starting position within the simulation area based on a random seed ( $X \times Y$ ).

It selects a location at random selected from the region to serve as the next way-point (W).

Then, it starts driving in the direction of the destination W at a speed that is randomly selected from the range of $V_{min}$ and $V_{max}$ .

After reaching destination W, the node pauses for a time randomly chosen $T_{p}$ seconds.

Then it repeats steps (2) to (4) until the simulation time $T_{max}$ finishes.

The RWP mobility model is based on GIS and has been designed to be applied to VANETs. It restricts the movement of vehicles to the streets that are indicated by the map data for real cities. Additionally, it limits the mobility of vehicles based on the vehicular congestion situation and simplifies traffic control methods. The runtime and memory usage of the mobility model are moderate and the model scales well with the size of the simulated network. We considered a GIS-based RWP mobility model to control the movement of vehicles.

4. Description of CT-based MS method and its evaluation

In this section, we present a CT-based MS method for V2V and V2I communication.

4.1. Overview of CT-based method

In [11], we have proposed CT-based MS method. The proposed method works together with typical DTN protocols such as Direct Delivery, SpW and Epidemic. MS method use three control packets to manage for delivering the bundle message (see Table 3).

Table 3
Control packets information for CT-based MS method

Packet Brief description of the packet role

MS-HELLO RSU sends a MS-HELLO to vehicles

ACK When vehicle receives MS-HELLO again from the same RSU, the vehicle replies the ACK

MS-Request When RSU receives the ACK, RSU sends a MS-Request to the vehicle

Packet	Brief description of the packet role
MS-HELLO	RSU sends a MS-HELLO to vehicles
ACK	When vehicle receives MS-HELLO again from the same RSU, the vehicle replies the ACK
MS-Request	When RSU receives the ACK, RSU sends a MS-Request to the vehicle

Fig. 3.

Sequence chart of CT-based MS method.

The RSUs send MS-HELLO message to vehicles at each interval. In the case when a vehicle, let say vehicle B, receives two messages from vehicle A, it sends the messages to other vehicles except vehicle A, as shown in Fig. 3. When vehicle B receives MS-HELLO message again from the same RSU, vehicle B sends the Acknowledgment (ACK) to the RSU. Then, RSU sends to vehicle B the MS-Request. When the RSU receives the ACK from some vehicles, the RSU knows the number of vehicles close to RSU. When vehicle B receives the MS-Request, it does not send the bundle messages to other vehicles, but keeps them in the storage. Thus, the volume of bundle traffic can be decreased.

4.2. Evaluation of CT-based MS method

We evaluated by simulation the network performance of a Non-Line of Sight (NLoS) scenario by sending the bundle messages for three DTN protocols. The NLoS communication environment is established by placing RSUs across building obstructions. There have been a total of forty iterations for each simulation.

Figure 4 shows the simulation results of PDR for different DTN protocols and RSUs. When the number of RSUs is zero (see the red line in Fig. 4), it indicates that RSUs are not constructed in grid road environments. This is the case for Conventional Epidemic, which does not use the CT-based MS method.

Fig. 4.

Results of PDR for different protocols.

For Direct Delivery (see Fig. 4(a)), we can see that results of PDR for this scenario are the same for different number of RSUs, because the start-point sends the bundle message to end-points directly. For Spray and Wait (see Fig. 4(b)), the results of PDR decreases with increasing the number of RSUs in the grid road environment. For Epidemic (see Fig. 4(c)), when the number of RSUs are more than 8, the PDR is higher than other DTN protocols, because RSU sends frequently MS-Request to vehicles. The difference of PDR is smaller than SpW with the increase of number of RSUs. In addition, when the number of RSUs is 16, the results of PDR is the best compared with other cases.

Fig. 5.

Results of storage usage for different protocols.

Fig. 6.

Results of mean E2E delay for different protocols.

Figure 5 shows the simulation results of the storage usage for different DTN protocols and RSUs. The bundle protocol starts to send bundle messages after 10 seconds. In Fig. 5(a), during the first transmission phase the mean storage usage increases proportionally because the number of stored bundle messages increases. After this phase, the storage usage increases slowly. In addition, the difference of the performance between different number of RSUs is very small. In Fig. 5(b), we found that mean storage usage is higher than the results of conventional Epidemic (RSU = 0). But, the storage usage is decreased when the number of RSUs reached twenty.

Figure 6 shows the simulation results of mean End-to-End (E2E) delay. It should be noted that only the data that successively reaches the endpoints are measured for the delay. The conventional method has a longer mean E2E delay than our proposed MS method, unless the number of RSUs is twenty. This is due to the proximity between the starting point and the RSUs. When many vehicles switch to MS, they do not transmit the bundle message to other vehicles.

5. Description of AT-based MS methods and performance evaluation

5.1. Overview of AT-based methods

Considering the evaluation results of CT-based MS method, we found that a good performance is achieved by combining CT-based method and Epidemic. Therefore, for two proposed AT-based MS methods we consider only the combination with Epidemic. These methods do not require the use of RSUs and instead implement suppression control based on the information obtained from neighboring nodes. This helps to decrease the overhead while support the mobility. We present the characteristics of the proposed AT-based MS methods in Table 4.

Table 4
Comparison of MS methods

Method Type Target Character Suppression time Threshold

MS Method CT RSU Forced message suppression Infinite Fixed

MSC AT Vehicle Decrease duplicated messages and support mobility Dynamic Fixed

MSC-ATh AT Vehicle Improve the overhead and mobility Dynamic Adaptive

Method	Type	Target	Character	Suppression time	Threshold
MS Method	CT	RSU	Forced message suppression	Infinite	Fixed
MSC	AT	Vehicle	Decrease duplicated messages and support mobility	Dynamic	Fixed
MSC-ATh	AT	Vehicle	Improve the overhead and mobility	Dynamic	Adaptive

5.2. MSC

The MSC runs message suppression procedures without V2I communication. It works together with SpW, MaxProp and Epidemic protocols. Each vehicle has a MSC, which is embedded in on-board unit of the vehicle. The MSC has active or inactive states. We present the flowchart of MSC algorithms for two states of vehicles in Fig. 7. About 10% of MSCs in the network are activated and send MS messages (such as MS-HELLO, MS-ACK and MS-REQUEST) to adjacent vehicles. These vehicles are called MSCVs. When MSCV receives HELLO message from other vehicles, the MSCV stores the bundle ID and source node ID in the memory (see Fig. 7(a)). In addition, MSCV detects the number of neighboring vehicles N to calculate the duration of message suppression.

Fig. 7.

Flowchart of MSC for MSCV and regular vehicles.

Fig. 8.

Sequence chart of MSC.

As shown in Fig. 8, the MSCV sends MS-HELLO packet to vehicles in intervals of 1 second. For example, when vehicle receives the MS-HELLO from MSCV, it sends a MS-ACK to the MSCV. When the MSCV receives the MS-ACK from a vehicle, the MSCV calculates Possession Rate of the Bundle ID ( ${PR}_{Bundle ID}$ ). The formula of ${PR}_{Bundle ID}$ is shown in Eq. (1). $\begin{array}{rcl} (1) & {PR}_{BundleID} = \frac{Number of detected same Bundle IDs}{N} \end{array}$ If ${PR}_{Bundle ID} ⩾ MST$ , the MSCV sends MS-REQUEST to the vehicle. The MS Threshold (MST) is the threshold value. We set the MST value to 1. Then, we used following steps to calculates the MS time for each MS-REQUEST. Before MSCV sends MS-REQUEST, the MSCV calculates the MS Time as shown in Eq. (2). $\begin{array}{rcl} (2) & {MS Time}_{Bundle ID} = t + ({MS Time}_{max} \times {PR}_{Bundle ID}) \end{array}$ In Eq. (2), t indicates the current time. The ${MS Time}_{max}$ is considered three hundred seconds. ${MS Time}_{Bundle ID}$ will be used in MS-REQUEST to suppress the bundle message. After that, the vehicle do not send the bundle message to other vehicles until the ${MS Time}_{Bundle ID}$ will be expired. As an exception, when vehicle moves to near end-points, the vehicle sends the bundle message to end-point even if ${MS Time}_{Bundle ID}$ is not expired.

The stored N will be reset every 2 seconds. In order to reduce the storage usage in the network, we use three addition packets as shown in Table 5. When a vehicle receives HELLO packet from other vehicles, the other vehicles send REQUEST packet to the vehicle. Then, if ${MS Time}_{Bundle ID}$ is less or equal with t, the vehicle sends the bundle message to other vehicles. When vehicle receives bundle message, the vehicle stores the bundle message in the memory. In this way, the amount of bundle traffic can be reduced.

Table 5

Control packets information for MSC and MSC-ATh

Packet	Brief description of the role
MS-HELLO	It is periodically broadcast, but it contains only the header information
MS-ACK	When vehicle receives the MS-HELLO from MSCV, the vehicle replies the MS-ACK
MS-REQUEST	When MSCV receives the MS-ACK, MSCV sends a MS-REQUEST to the vehicle
MS-REQUEST	It contains the details on the list of bundle messages in the storage

5.3. MSC-ATh

In the conventional MSC, we considered a fixed threshold value of MST. For instance, when ${PR}_{BundleID} ⩾ 1$ , conventional method does not replicate the bundle message for a bundle ID. We present an adaptive method to increase the network performance in V2V communication. We propose a MSC-ATh and present the performance evaluation without V2I communication. We show the flowchart of proposed MSC-ATh algorithm in Fig. 9.

Fig. 9.

Flowchart of MSC-ATh.

When vehicle receives MS-HELLO from MSCV, it sends a MS-ACK to the MSCV. In this case, vehicle receives many MS-HELLOs in a short time, thus increasing processing overhead. For this reason, we proposed an Adaptive Threshold (ATh) for MSC to solve the problem. The method increases efficiency by sending fewer reply packets than the conventional MSC. The ATh has a timer as shown is Fig. 9. $\begin{array}{rcl} (3) & ATh = \frac{SC}{N}; N > 1 \end{array}$ In Eq. (3), the SC indicates suppression coefficient. We consider the SC values from 2 to 8. We designed that the threshold for message suppression to increase with the increase of the number of adjacent vehicles. When N is less or equal to one, the MSC-ATh does not suppress the bundle message. When N and SC are the same, the threshold is the same compared with conventional MSC. The N value increases with the increase of the message suppression. When ${PR}_{BundleID} ⩾ ATh$ , MSCV sends MS-REQUEST to the vehicle.

5.4. Evaluation of MSC and MSC-ATh methods

We evaluate the proposed MSC and MSC-ATh and compare with conventional Epidemic. For evaluation, we did not consider ant-packet recovery scheme. For evaluation, we consider three different SCs and following parameters: PDR, delay, duplicated bundles, control packets and storage usage for different number of vehicles.

Fig. 10.

Performance comparison by protocol for each evaluation metric.

5.4.1. Comparative assessment by different protocols

We evaluate the network performance for different protocols as shown in Fig. 10 and Fig. 11. The value of SC is set to 2.

In Fig. 10(a), we present the simulation results of PDR for different protocols. For different vehicles, we observe that all bundle messages reach the end-point by the end of the simulation time. In Fig. 10(b), in case of the sparse network, there is a large difference in delay time due to message suppression. We found that Epidemic, which does not process suppression requests and recovery, sends messages to their target in a timely manner. The difference in the delay performance is decreased after hundred vehicles in this scenario. Delay increases with increasing of number of vehicles. In Fig. 10(c), we show the simulation results of duplicated bundles for different protocols. We observe that both MSC and MSC-ATh perform well and they have better performance than Epidemic. In Fig. 10(d), we show the simulation results of generated control packets in the network for different protocols. With the new MS-HELLO, MS-ACK, and MS-REQUEST messages added to the proposed MSC and MSC-ATh methods, the number of control packets in the network is increased. However, the size of each control packet is small and had no effect on PDR performance.

In Fig. 11, we show the simulation results of storage usage for different protocols. The variability of storage usage improves with the increasing number of vehicles. The MSC-ATh has better performance compared with other protocols.

Fig. 11.

Storage usage for different protocols.

5.4.2. Comparative assessment of MSC-ATh for SC parameter

We evaluate the network performance of MSC-ATh for different SCs (2, 4 and 8) as shown in Fig. 12 and Fig. 13. We use SC for the MSC-ATh as shown in Eq. (3).

In Fig. 12(a), we present the simulation results of PDR for different SCs. The PDR reaches hundred percent by the end of the simulation. In Fig. 12(b), we show the simulation results of delay for different SCs. The delay performance improves with increasing the SCs and number of vehicles. When SC increases above 4, we can see an improvement in delay performance compared to MSC (see Fig. 10(b) and Fig. 12(b)). In Fig. 12(c), we show the simulation results of duplicated bundles for different SCs. The duplicated bundles increase with the increasing the number of vehicles. Also, for 50 and 150 vehicles, the duplicated bundles increase with increasing the SCs. In Fig. 12(d), we show the simulation results of generated control packets in the network for different SCs. The control packets decreases with increasing SCs. The results show that by increasing SC the number of control packets and delay can be decreased, but the number of duplicated messages is increased, even in the case of sparse network.

In Fig. 13, we show the simulation results of storage usage for different SCs. When SC is increased, there is a slight increase in storage utilization. However, this increase is small compared with other protocols, which can contribute to storage reduction.

Fig. 12.

Comparison of the MSC-ATh performance for each metric based on the different SCs.

Fig. 13.

Storage usage of MSC-ATh performance for different SCs.

6. Conclusions

In this paper, we presented CT-based and AT-based MS methods for Vehicular DTN. For CT-based MS method, we used Direct Delivery, SpW and Epidemic protocols and different vehicles and RSUs. For AT-based MSC and MSC-ATh, we used conventional Epidemic and our Epidemic-based methods for different vehicles without RSUs. Then, we evaluated the MSC-ATh for different SCs. We compared MS method, MSC and MSC-ATh. From the evaluation results, we conclude as follows.

The conventional method has a longer mean E2E delay than our proposed MS method, unless the number of RSUs is twenty. This is due to the proximity between the starting point and the RSUs. When many vehicles switch to MS, they do not transmit the bundle messages to other vehicles.

For Epidemic, when the number of RSUs is more than 8, the PDR is higher than Direct Delivery and SpW protocols, because RSUs send frequently MS-Request to vehicles. The difference of PDR is smaller than SpW when the number of RSUs increases. In addition, when the number of RSUs is 16, the results of PDR is the best compared with other cases.

The variability of storage usage improves with the increasing number of vehicles. The MSC-ATh has better performance compared with other protocols.

The delay performance of MSC-ATh improves with increasing the SCs and number of vehicles.

By increasing SC, the number of control packets and delay can be decreased, but the number of duplicated messages is increased, even in the case of sparse network.

In future work, we would like to evaluate other MS methods and consider other parameters. We also would like to consider the anti-packet recovery method to manage the bundle messages in VANETs.

Conflict of interest

The authors have no conflict of interest to report.

References

Barroca ,

Grilo and

P.R.

Pereira , Improving message delivery in UAV-based delay tolerant networks, in: Proceedings of the 16th International Conference on Intelligent Transportation Systems Telecommunications (ITST-2018), 2018, pp. 1–7. doi:10.1109/ITST.2018.8566956.

Burgess ,

Gallagher ,

Jensen and

B.N.

Levine , MaxProp: Routing for vehicle-based disruption-tolerant networks, in: Proceedings of the 25th IEEE International Conference on Computer Communications (IEEE INFOCOM-2006), 2006, pp. 1–11.

Cao ,

Jiang ,

Kaiwartya ,

Sun ,

Zhou and

Wang , Toward pre-empted EV charging recommendation through V2V-based reservation system, IEEE Transactions on Systems, Man, and Cybernetics: Systems 51(5) (2021), 3026–3039. doi:10.1109/TSMC.2019.2917149.

Cerf ,

Burleigh ,

Hooke ,

Torgerson ,

Durst ,

Scott ,

Fall and

Weiss , Delay-Tolerant Networking Architecture, 2007.

M.C.

Chuah and

W.-B.

Ma , Integrated buffer and route management in a DTN with message ferry, in: Proceedings of the IEEE Military Communications Conference (MILCOM-2006), 2006, pp. 1–7. doi:10.1109/MILCOM.2006.302288.

Davarian ,

Asmar ,

Angert ,

Baker ,

Gao ,

Hodges ,

Israel ,

Landau ,

Lay ,

Torgerson and

Walsh , Improving small satellite communications and tracking in deep space – a review of the existing systems and technologies with recommendations for improvement. Part II: Small satellite navigation, proximity links, and Communications Link Science, IEEE Aerospace and Electronic Systems Magazine 35(7) (2020), 26–40. doi:10.1109/MAES.2020.2975260.

Fall , A delay-tolerant network architecture for challenged internets, in: Proceedings of the International Conference on Applications, Technologies, Architectures, and Protocols for Computer Communications, SIGCOMM ’03, 2003, pp. 27–34.

J.A.

Fraire ,

Feldmann and

S.C.

Burleigh , Benefits and challenges of cross-linked ring road satellite networks: A case study, in: Proceedings of the IEEE International Conference on Communications (ICC-2017), 2017, pp. 1–7. doi:10.1109/ICC.2017.7996778.

Henkel and

T.X.

Brown , Delay-tolerant communication using mobile robotic helper nodes, in: Proceedings of the 6th International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks and Workshops 2008, 2008, pp. 657–666. doi:10.1109/WIOPT.2008.4586155.

10.

Henmi and

Koyama , Hybrid type DTN routing protocol considering storage capacity, in: Proceedings of the 8th International Conference on Emerging Internet, Data and Web Technologies (EIDWT 2020), 2020, pp. 491–502.

11.

Honda ,

Ishikawa ,

Ikeda and

Barolli , A message suppression method for vehicular delay tolerant networking, in: Proceedings of the 5th International Workshop on Methods, Analysis and Protocols for Wireless Communication (MAPWC-2014), 2014, pp. 351–356.

12.

Hu ,

Tian ,

Zhang ,

Chen ,

Yu and

Ma , Restricted epidemic routing method in large-scale delay-tolerant networks, in: Proceedings of the IEEE 21st International Conference on Communication Technology (ICCT-2021), 2021, pp. 380–386. doi:10.1109/ICCT52962.2021.9657947.

13.

Ikeda ,

Ishikawa ,

Honda and

Barolli , Impact of location of road-side units considering message suppression method for vehicular-DTN, in: Proceedings of the 9th International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing (IMIS-2015), 2015, pp. 153–159.

14.

Iranmanesh ,

Raad ,

M.S.

Raheel ,

Tubbal and

Jan , Novel DTN mobility-driven routing in autonomous drone logistics networks, IEEE Access 8 (2020), 13661–13673. doi:10.1109/ACCESS.2019.2959275.

15.

Jahanian and

K.K.

Ramakrishnan , CoNICE: Consensus in intermittently-connected environments by exploiting naming with application to emergency response, IEEE/ACM Transactions on Networking (2022), 1–14. doi:10.1109/TNET.2022.3156101.

16.

Lent , Enabling cognitive bundle routing in NASA’s high rate DTN, in: Proceedings of the 2 International Wireless Communications and Mobile Computing (IWCMC-2022), 2022, pp. 1323–1328. doi:10.1109/IWCMC55113.2022.9824158.

17.

Liang ,

Gao ,

J.H.

Nguyen ,

M.F.

Orpilla and

Yu , Internet of things data collection using unmanned aerial vehicles in infrastructure free environments, IEEE Access 8 (2020), 3932–3944. doi:10.1109/ACCESS.2019.2962323.

18.

Ramanathan ,

Hansen ,

Basu ,

R.R.

Hain and

Krishnan , Prioritized epidemic routing for opportunistic networks, in: Proceedings of the 1st International MobiSys Workshop on Mobile Opportunistic Networking (MobiOpp ’07), 2007, pp. 62–66. doi:10.1145/1247694.1247707.

19.

Rüsch ,

Schürmann ,

Kapitza and

Wolf , Forward secure delay-tolerant networking, in: Proceedings of the 12th Workshop on Challenged Networks (CHANTS-2017), 2017, pp. 7–12. doi:10.1145/3124087.3124094.

20.

Sagayama ,

Matsuo and

Ohsaki , On the effect of communication link heterogeneity on content delivery delay in information-centric delay tolerant networks, in: Proceedings of the IEEE 46th Annual Computers, Software, and Applications Conference (COMPSAC-2022), 2022, pp. 1044–1049. doi:10.1109/COMPSAC54236.2022.00163.

21.

Scenargie, Space-time Engineering, LLC, http://www.spacetime-eng.com/.

22.

Scott and

Burleigh , Bundle Protocol Specification, 2007.

23.

Short ,

Hylton ,

Cleveland ,

Moy ,

Cardona ,

Green ,

Curry ,

Mallery ,

Bainbridge and

Memon , Sheaf theoretic models for routing in delay tolerant networks, in: Proceedings of the IEEE Aerospace Conference (AERO-2022), 2022, pp. 1–19. doi:10.1109/AERO53065.2022.9843504.

24.

Solpico ,

M.I.

Tan ,

E.J.

Manalansan ,

F.A.

Zagala ,

J.A.

Leceta ,

D.F.

Lanuza ,

Bernal ,

R.D.

Ramos ,

R.J.

Villareal ,

X.M.

Cruz ,

J.A.

dela Cruz ,

D.J.

Lagazo ,

J.L.

Honrado ,

Abrajano ,

N.J.

Libatique and

Tangonan , Application of the V-HUB standard using LoRa beacons, mobile cloud, UAVs, and DTN for disaster-resilient communications, in: Proceedings of the IEEE Global Humanitarian Technology Conference (GHTC-2019), 2019, pp. 1–8.

25.

Spyropoulos ,

Psounis and

C.S.

Raghavendra , Spray and wait: An efficient routing scheme for intermittently connected mobile networks, in: Proceedings of the ACM SIGCOMM Workshop on Delay-Tolerant Networking 2005 (WDTN ’05), 2005, pp. 252–259. doi:10.1145/1080139.1080143.

26.

Sugihara and

Hayashibara , Message delivery of nomadic Lévy walk based message ferry routing in delay tolerant networks, in: Proceedings of the 36th International Conference on Advanced Information Networking and Applications (AINA-2022), Lecture Notes in Networks and Systems, Vol. 449, 2022, pp. 259–270. ISBN 978-3-030-99584-3. doi:10.1007/978-3-030-99584-3_23.

27.

Tada ,

Ikeda and

Barolli , Performance evaluation of a message relaying method for resilient disaster networks, in: Proceedings of the 15th International Conference on Broadband and Wireless Computing, Communication and Applications (BWCCA-2020), 2020, pp. 1–10.

28.

Uchimura ,

Azuma ,

Tada ,

Ikeda and

Barolli , An adaptive anti-packet recovery method for vehicular DTN considering message possession rate, in: Proceedings of the 35th International Conference on Advanced Information Networking and Applications (AINA-2021), Vol. 1, 2021, pp. 92–101. ISBN 978-3-030-75100-5. doi:10.1007/978-3-030-75100-5_9.

29.

Vahdat and

Becker , Epidemic Routing for Partially-Connected Ad Hoc Networks, Technical Report, Duke University, 2000.

30.

Wyatt ,

Burleigh ,

Jones ,

Torgerson and

Wissler , Disruption tolerant networking flight validation experiment on NASA’s EPOXI mission, in: Proceedings of the 1st International Conference on Advances in Satellite and Space Communications (SPACOMM-2009), 2009, pp. 187–196.

31.

Xu ,

Zhang ,

Pang ,

Kang and

Su , Design and implementation of a transport protocol with network coding for delay tolerant underwater acoustic sensor networks, in: Proceedings of the OCEANS 2022, 2022, pp. 1–4. doi:10.1109/OCEANSChennai45887.2022.9775134.

32.

Yasmeen ,

Huda ,

Yamada and

Borcea , Ferry access points and sticky transfers: Improving communication in ferry-assisted DTNs, in: Proceedings of the IEEE International Symposium on a World of Wireless, Mobile and Multimedia Networks (WoWMoM-2012), 2012, pp. 1–7. doi:10.1109/WoWMoM.2012.6263746.

33.

Yoshino ,

Nakasaki ,

Ikeda and

Barolli , A threshold-based adaptive method for message suppression controller in vehicular DTNs, in: Proceedings of the 13th International Confernce on Broad-Band Wireless Computing, Communication and Applications (BWCCA-2018), 2018, pp. 517–524.

34.

Zhao ,

Ammar and

Zegura , Controlling the mobility of multiple data transport ferries in a delay-tolerant network, in: Proceedings IEEE 24th Annual Joint Conference of the IEEE Computer and Communications Societies, Vol. 2, 2005, pp. 1407–1418. doi:10.1109/INFCOM.2005.1498365.

35.

Zhao and

M.H.

Ammar , Message ferrying: Proactive routing in highly-partitioned wireless ad hoc networks, in: The Ninth IEEE Workshop on Future Trends of Distributed Computing Systems, 2003. FTDCS 2003. Proceedings., 2003, pp. 308–314. doi:10.1109/FTDCS.2003.1204352.

Performance evaluation of contact-time based and adaptive-timer based message suppression methods for inter-vehicle communication in vehicular DTN

Abstract

Keywords

1. Introduction

2. Related work

2.1. DTN

2.2. Inter-Vehicle Communication based on V2V and V2I

3. Simulation modeling

3.1. Setup for CT-based MS method

4. Description of CT-based MS method and its evaluation

4.1. Overview of CT-based method

5.1. Overview of AT-based methods

Table 4 Comparison of MS methods Method Type Target Character Suppression time Threshold MS Method CT RSU Forced message suppression Infinite Fixed MSC AT Vehicle Decrease duplicated messages and support mobility Dynamic Fixed MSC-ATh AT Vehicle Improve the overhead and mobility Dynamic Adaptive

Conflict of interest

References