Density-Aware Delay-Tolerant Interest Forwarding in

Density-Aware Delay-Tolerant Interest Forwarding in Vehicular Named Data Networking Meng Kuai∗ , Xiaoyan Hong∗ , Qiangyuan Yu † of Computer Science, The University of Alabama, Tuscaloosa, AL 35487, USA † College of Computer Science and Technology, Jilin University, Changchun 130012, China [email protected], [email protected], [email protected]

∗ Department

Abstract—Named Data Networking (NDN) has been considered as a promising networking architecture for Vehicular Ad-Hoc Networks (VANETs). However, Interest forwarding in NDN suffers severe issues in vehicular environment. Broadcast storm results in much packet loss and huge transmission overhead. Also, link disconnection caused by highly dynamic topology leads to low packet delivery ratio. On the other hand, traffic data are playing significant roles in VANETs since they are essential in varieties of Intelligent Transportation System (ITS) applications. Thus, an efficient NDN forwarding strategy using geographical characteristics to retrieve traffic data is urgently required. In this paper, we propose Density-Aware Delay-Tolerant (DADT) Interest forwarding strategy to retrieve traffic data in vehicular NDN with the purpose of improving packet delivery ratio. DADT specifically addresses data retrieval during network disruptions using Delay Tolerant Networking (DTN). It makes retransmission decision based on directional network density. Also, DADT mitigates broadcast storm by using rebroadcast deferring timer. We compared DADT against other strategies through simulation and the results show that it can achieve higher satisfaction ratio while keeping low transmission overhead.

I. I NTRODUCTION Vehicular Ad-Hoc Networks (VANETs) have attracted much attention from academic researchers to automotive manufactures for over a decade because they support varieties of Intelligent Transportation System (ITS) applications including safe driving, traffic flow controllers and navigation assistance. Most of the applications require traffic data. Therefore, an efficient approach to retrieve traffic data in VANETs is required. However, traditional host-based network architecture limits the message dissemination in VANETs. Named Data Networking (NDN) [5][12] has been proposed and has become a promising architecture for the next generation of networking. In NDN, all communications are based on data name, instead of host-tohost connections. To request data, consumers should send an Interest packet with a data name to the network. Any provider who has the name-matching data will send it back. Due to its connection-free feature, as well as multicast and in-path caching, NDN has been recognized as an attractive solution for VANETs [8][3]. Several issues have appeared while building VANETs to support NDN architecture because of the wireless communication medium and dynamic network topology. On one hand, Forwarding Information Base (FIB) seems to be unfeasible in vehicular environment because it is difficult to maintain in such highly dynamic topology. Instead of forwarding Interest

based on FIB, each forwarder simply rebroadcasts Interest to the next hop, which may lead to severe broadcast storm. On the other hand, network connectivity in VANETs is highly unstable and can vary rapidly. Delay Tolerant Networking (DTN) is a common approach to improve robustness during network disruptions. In DTN, packets are stored and carried at nodes until there are chances to forward them. How to enhance NDN to support DTN has not been studied yet. In this paper, we propose Density-Aware Delay-Tolerant (DADT) Interest forwarding strategy to retrieve traffic data in vehicular NDN with the purpose of improving packet delivery ratio. To be specific, each node opportunistically maintains a neighbor list by overhearing all incoming packets in-the-air and makes retransmission decision based on detected directional network density. To mitigate broadcast storm, DADT uses a new “deferring timer” that is calculated based on the locations of the nodes and data sources. The times and the locations are the main characteristics of the vehicular networks used in the calculations because the transportation traffic data is bounded with temporal and geographical coordinate information. Most of the applications that are related to realtime traffic information can benefit from DADT. We completely implemented DADT through simulation and evaluated its performance by comparing against other strategies. After analyzing the results, we find that DADT achieves higher satisfaction ratio while keeping low transmission overhead. The rest of the paper is organized as follows. We discuss related work in section II. Section III introduces our proposed forwarding strategy. And section IV shows our simulation configuration and performance evaluation results. At last, section V concludes the paper. II. R ELATED W ORK NDN data communication model brings potential advantages to VANETs. It appears to be more robust to overcome the mobility and intermittent connectivity challenges in VANETs. Recently, several applications of Information-Centric Networking (ICN) to VANETs have been published. Amadeo et al. [2] focused on the design and evaluation of the NDN forwarding plane in Mobile Ad-Hoc Networks (MANETs). It proposed provider-blind forwarding strategy and provideraware strategy which use defer timer and overhearing to deal with flooding. Grassi et al. proposed Navigo [4], which is a location based packet forwarding mechanism for V-NDN

978-1-5090-1701-0/16/$31.00 ©2016 IEEE

[3]. It can guide Interests towards data producers using a specialized shortest path over the road topology by mapping data names to data locations and forwarding Interest packets along the best path. Yu et al. proposed an opportunistic geoinspired content based routing method [11], which utilizes the last encounter information of each node to infer the locations of content holders. With the last encounter information, the Interests can be geo-routed instead of being flooded to reduce the congestion level of the entire network and highest priority is always given to forwarders who have relatively recent location information. Wang et al. proposed and evaluated an application for V2V traffic information dissemination that leverages NDN for efficient Interest and data broadcasting [9]. It employs a set of timers to coordinate transmissions and minimize packets collisions on the shared wireless medium. Neighborhood-Aware Interest Forwarding (NAIF) [10] has been proposed by Yu et al.. It adjusts forwarding rate dynamically based on forwarding statistic information it overhears locally and estimates missed rate. In this way, each node reduces forwarding overhead by sharing workload with its neighbors. Lu et al. proposed a DTN scheme to retrieve content in MANETs using social centrality [7]. Most related works mitigate broadcast storm using deferring timer and overhearing. But these schemes had not specially dealt with DTN cases in VANETs. Our paper aims to design a scheme which considers rebroadcast and retransmission together, also including the DTN case which is important for intermittent VANETs.

A. Rebroadcast In the current V-NDN, FIB is not used in the Interest forwarding process, each candidate forwarder simply rebroadcasts the pending Interest, which will introduce several issues. First of all, packet collisions will occur frequently if neighboring nodes rebroadcast Interest around the same time. Secondly, blind broadcast will make Interest propagate to all directions and flood the network. Last but not the least, making every node forwarding Interest arbitrarily will bring much unnecessary transmission overhead. To deal with those issues, DADT takes advantage of the rebroadcast deferring timer that sets up a waiting period for each forwarder before their actual transmission. It can reduce the probabilities of packet collision and significant benefit is selective forwarding. Among several neighboring nodes that receive the same Interest, each node sets up deferring timer based on its forwarding priority and adds the Interest to its pending transmission queue. The node with higher forwarding priority will have shorter waiting timer. During the waiting period of each node, if it overhears an incoming Interest/data packet that has the same name with the previously scheduled transmitted Interest, it means that the Interest has been forwarded by other node that has higher forwarding priority than the node itself. So it cancels the scheduled transmission for the Interest. Eventually, only parts of nodes with higher forwarding priority actually forward the Interest.

III. F ORWARDING S TRATEGY D ESIGN In this section, we introduce our proposed Density-Aware Delay-Tolerant (DADT) Interest forwarding strategy. Following the convention, we acknowledge that each vehicle is equipped with GPS device and is able to obtain its current geographic location. For the transportation traffic data to be retrieved in NDN, each data has a name that attaches the geographical coordinate information. In our implementation, we use the name structure of /(latitude,longitude)/sequence to represent the traffic data. In addition, whenever an Interest or data packet is sent from the network interface, a node will attach its ID and current geographic location to the packet. Each node maintains a neighbor list, which is updated whenever it overhears a packet in-the-air. The neighbor list records the timestamp and the last hop’s ID and location. Practically, the location information can be attached in the periodic DSRC safety beacons and will not bring additional overhead. In our implementation, every second node sends out a special beacon Interest packet with name /BEACON/, which cannot be forwarded by other nodes and expires in one second. Our forwarding strategy includes two communication phases: rebroadcast and retransmission. Rebroadcast occurs immediately after a node receives an Interest packet. Retransmission happens when a node stores the Interest packet for a while after forwarding and retransmits it.



Z



Eϭ

ɲϭ ɽ



ɲϮ

EϮ

Fig. 1. Forwarder Priority

We considered two factors indicating forwarding priority when assigning rebroadcast deferring timers. On one hand, the nodes farther away from the last hop should have higher forwarding priority so that the Interest can be propagated faster with fewer hops. On the other hand, the nodes closer to data location should have higher forwarding priority thus the Interest could have more chance to be forwarded to its desired area. To consider these two factors, we make use of the reference node, which is defined as the nearest possible geocoordinate to the data location within the transmission range of last hop [11]. For example, in Fig. 1, node A receives an Interest packet coming from node B and the name indicates data location is D. So R is the reference node. In our design,

a node closer to the reference node has shorter rebroadcast deferring timer because of higher forwarding priority. We use the following equation to compute the timer: 𝑇𝑟𝑒𝑏𝑟𝑜𝑎𝑑𝑐𝑎𝑠𝑡

𝐷𝑅 = 𝑇𝑚𝑎𝑥 + 𝑇𝑟𝑎𝑛𝑑𝑜𝑚 2𝐷𝑚𝑎𝑥

(1)

where 𝑇𝑚𝑎𝑥 is the maximum waiting time, 𝐷𝑚𝑎𝑥 is the transmission range and 𝐷𝑅 indicates the distance to reference 𝐷𝑅 node. The value range of is from 0 to 1 so that the 2𝐷𝑚𝑎𝑥 range of waiting time is from 0 to 𝑇𝑚𝑎𝑥 . To avoid packet collision when two calculated timers are close, a random timer 𝑇𝑟𝑎𝑛𝑑𝑜𝑚 is appended to the equation. B. Retransmission VANETs can be partitioned frequently. In the case of network interruption, NDN suffers low delivery ratio. DTN is a common approach to address that problem, in which node store-and-carry the packet and retransmits it after a while. There are several questions when designing DADT: (1) is the retransmission always necessary? (2) when to retransmit for better performance? DADT aims to improve packet delivery ratio while trying to keep transmission overhead low. Also, the DTN response delay should be acceptable. In DADT, after forwarding the Interest, each node will add it to a pending retransmission queue. The node will not retransmit the Interest until it detects an appropriate neighbor with forwarding capability. It means the packet may be forwarded to the desired areas through that neighbor. If the node overhears an Interest with the same name that sent from another node with stronger forwarding capability, it will delete the Interest from its pending retransmission queue. Because this Interest has been successfully forwarded in this area. The retransmission will also be canceled if the node overhears a data packet with the same name since the Interest has already been satisfied. In Fig. 1, to express the potential forwarding capability of nodes, we define a sector area between A and D with an angle of 𝜃. Clearly, the nodes insides this area will have more probabilities to forward the Interest towards its desired 𝑖 , location. In our design, we define spatial priority 𝑃𝑠𝑝𝑎𝑡𝑖𝑎𝑙 which measures how close the neighbor is to the shortest line between the current node and the data location. It is calculated according to the neighbor list. Assume node A maintains an up-to-date neighbor list 𝑁 . For each 𝑁𝑖 in 𝑁 , the spatial 𝑖 can be expressed as follows: priority 𝑃𝑠𝑝𝑎𝑡𝑖𝑎𝑙 𝐷2 (𝐴, 𝑁𝑖 ) + 𝐷2 (𝐴, 𝐷) − 𝐷2 (𝑁𝑖 , 𝐷) 𝛼𝑖 = cos−1 2𝐷(𝐴, 𝑁𝑖 )𝐷(𝐴, 𝐷) ⎧  ⎨ 𝜃 − 2𝛼𝑖 , 2𝛼𝑖 < 𝜃 𝑖 𝑃𝑠𝑝𝑎𝑡𝑖𝑎𝑙 = 𝜃  ⎩0, 2𝛼𝑖 >= 𝜃

(2)

(3)

where function D calculates the distance between two nodes. 𝑖 indicates the spatial priority of each In equation (3), 𝑃𝑠𝑝𝑎𝑡𝑖𝑎𝑙 neighbor.

The summarized neighbor priority can be presented as: 𝑃𝑛𝑒𝑖𝑔ℎ𝑏𝑜𝑟 =

∣𝑁 ∣ ∑

𝑖 𝑃𝑠𝑝𝑎𝑡𝑖𝑎𝑙

(4)

𝑖=1

We use 𝑃𝑛𝑒𝑖𝑔ℎ𝑏𝑜𝑟 as a threshold to make retransmission decisions. After updating the neighbor list, node checks the 𝑃𝑛𝑒𝑖𝑔ℎ𝑏𝑜𝑟 for each packet in retransmission queue. If 𝑃𝑛𝑒𝑖𝑔ℎ𝑏𝑜𝑟 > 0, node retransmits the packet immediately. Before retransmitting, if a node overhears the same Interest from another node, it will check if the Interest’s last hop is in the sector area. If yes, it means the packet has been forwarded by nodes with stronger forwarding capability, so it cancels its own retransmission. C. Overall Workflow hƉĚĂƚĞ EĞŝŐŚďŽƌ>ŝƐƚ

zĞƐ

/ŶĐŽŵŝŶŐ /ŶƚĞƌĞƐƚWĂĐŬĞƚ

/ƐďĞĂĐŽŶ /ŶƚĞƌĞƐƚ͍ EŽ

WŶĞŝŐŚďŽƌхϬ

/ƐŝŶ/ŶƚĞƌĞƐƚddž YƵĞƵĞ͍

zĞƐ

ĞůĞƚĞĨƌŽŵ ddžYƵĞƵĞ

zĞƐ >ĂƐƚŚŽƉŝŶ ƐĞĐƚŽƌĂƌĞĂ͍

ZĞƚdž

/ƐŝŶ/ŶƚĞƌĞƐƚ ZĞƚdžYƵĞƵĞ͍

ĞůĞƚĞĨƌŽŵ ZĞƚdžYƵĞƵĞ

EŽ

zĞƐ

/ƐŝŶĂƚĂddž YƵĞƵĞ͍

ĞůĞƚĞĨƌŽŵ ZĞƚdžYƵĞƵĞ

ƌŽƉ

ƌŽĂĚĐĂƐƚ ĂƚĂ

zĞƐ zĞƐ

zĞƐ

zĞƐ

EŽ /ƐŝŶW/d͍

ĚĚƚŽZĞƚdž YƵĞƵĞ

EŽ

ĚĚƚŽddžYƵĞƵĞ

/ƐŝŶ^͍

EŽ

^ĞƚƵƉƌĞďƌŽĂĚĐĂƐƚ ƚŝŵĞƌ

Fig. 2. DADT Interest Forwarding Strategy Workflow

Fig. 2 represents the overall workflow for our proposed Density-Aware Delay-Tolerant (DADT) Interest forwarding strategy. Each node maintains transmission and retransmission queues for Interest and transmission queue for data. Whenever node overhears a packet, it will check and update the queues before it checks Pending Interest Table (PIT) and Content Store (CS). Notice that if this Interest has a naming-matching data in the queues, it will be dropped because it has already been satisfied. For the data forwarding, we set up rebroadcast deferring times using the distance to the last hop [9]. IV. P ERFORMANCE E VALUATION In this section, we introduce our work on evaluating the proposed forwarding strategy, including the configuration of our simulation, the metrics we used to measure the performance and other strategies we compared with. Also, we present the performance results. A. Simulation Configuration Our simulation map is an urban area spanning about 1 × 1 km in San Francisco. We imported the map into SUMO [6] and generated a certain number of vehicles randomly distributed

In the following representation, we use 𝑁 to indicate the total number of issued Interest during the whole simulation, 𝑛 to indicate the number of satisfied Interest that consumer obtains the requested data through instant connection and 𝑚 to indicate the number of satisfied Interest that consumer obtains the requested data by DTN. We have run the simulation 30 times with different random seeds to obtain the average value for these metrics. 𝑛+𝑚 . This metric 1) Satisfaction ratio: it is defined as 𝑁 indicates the delivery ratio of Interest and data packet on application level. 2) Normalized Transmission overhead: it is the total number of transmitted Interest and data packets for each forwarded node per issued Interest. This metric can measure the transmission cost of network. We did not count beacon Interest into transmission overhead. ∑𝑚 𝑖=1 𝑡𝑖 , where 𝑡𝑖 3) Average DTN delay: it is defined as 𝑚 indicates the delay of each satisfied Interest by DTN. We compare the performance of DADT against three other strategies: (a) First Contact: node retransmits Interest when it encounters the first neighbor; (b) 0.5 Probability: node has 50% chance to retransmit Interest when it encounters each neighbor; (c) No retx: node does not retransmit Interest. Notice that all DTN strategies only retransmit Interest one time. For the parameters when calculating timers, we set 𝐷𝑚𝑎𝑥 as 150 𝑚𝑒𝑡𝑒𝑟𝑠, 𝑇𝑚𝑎𝑥 as 0.03 𝑠𝑒𝑐𝑜𝑛𝑑𝑠 and 𝜃 as 𝜋/2. We have also studied the impact of 𝜃 on the performance. For simulating the variety of topology density and Interest request load to study their influence on network performance,

C. Evaluation Results

Satisfication Ratio

0.7

0.55

DADT First Contact 0.5 Probability No retx

0.50 0.45 Satisfication Ratio

0.9 0.8

0.6 0.5

0.4

0.35 0.30 0.25

0.3 0.2 20

0.40

DADT First Contact 0.5 Probability No retx

0.20

30

40

50

Number of Cars

60

70

80

0.15 2

4

6

8 10 Number of Consumers

12

14

16

Fig. 3. Satisfaction Ratio

Fig. 3 shows the performance of satisfaction ratio. We observe that our proposed DADT has the highest satisfaction ratio among all strategies. Particularly in sparse topology, it achieves over 30% higher satisfaction ratio than other two DTN strategies. Even though DADT only retransmits Interest once, it achieves almost double satisfaction ratio of No retx. It is because DADT takes into account spatial priority when making retransmission decisions. We find that denser topology helps to increase satisfaction ratio. The gap size between the curves of DADT and No retx indicates those satisfactions that have been completed by retransmission using DTN, which does not change too much in different topology densities. It means that denser topologies introduce more chance for instant connections, but not for DTN. In addition, more consumers contribute to higher satisfaction ratio since more Interests can be satisfied by the data in cache. But the curves become relatively gentle when number of consumers is larger than 6, which means that in those cases, network topology becomes more critical factor to achieve higher satisfaction ratio.

10

DADT First Contact 0.5 Probability No retx

DADT First Contact 0.5 Probability No retx

Normalized Transmission Overhead

B. Evaluation Method

in our evaluation, we adjusted the number of cars and the number of consumers, respectively. The consumer number maintains 4 when the number of cars varies. The car number keeps 30 when the number of consumers changes.

Normalized Transmission Overhead

on map. During the 500-second simulation, each vehicle picks up a random point on the map as a destination and calculates a shortest path. Once it reaches the point, it will randomly choose the next destination and keep doing so until the end of simulation. We chose two Region of Interest (ROI) and deployed Road Side Unit (RSU) on them. RSU can detect nearby traffic information and generate data in response to Interest packets. In our simulation, the two RSUs are running producer applications and replying data upon Interest. A subset of vehicles are running consumer applications by issuing sequential Interest at a certain period (10 seconds). Every Interest is requesting traffic data generated by one of the RSUs and will expire in 50 seconds. Other vehicles are playing the role of forwarders to rebroadcast in instant connection and/or data mules to retransmit using DTN. This configuration is to mimic the situation in VANETs that vehicles periodically request sequential traffic information in its interested area. We implemented Density-Aware Delay-Tolerant (DADT) Interest forwarding strategy in a modified version of ndnSIM 2.0 [1]. The strategy module was added to NetDevice interface without making any changes the default NDN Forwarding Daemon (NFD). All vehicles and RSUs are equipped with Wi-Fi Ad-Hoc interfaces and installed with NDN stack.

4.5

4.0

8

3.5

6

3.0 2.5

4

2.0

2 0 10

1.5

20

30

40

50

Number of Cars

60

70

80

1.0 2

4

6

8

10

12

Number of Consumers

14

16

Fig. 4. Normalized Transmission Overhead

The performance of normalized transmission overhead is shown in Fig. 4. As observed, DTN strategies have a little larger transmission overhead than other two DTN strategies and No retx, which is due to the fact that it has higher satisfaction ratio thus Interest and data transmission overhead are larger. We find that this metric has a linear relation to the number of cars. Also, more consumers requesting the same

data will reduce the overhead because fewer Interests will be sent to the data source area. 18

14

16

Average DTN Delay (seconds)

Average DTN Delay (seconds)

18

DADT First Contact 0.5 Probability

16

V. C ONCLUSION

14

12

12

10

DADT First Contact 0.5 Probability

10

8 6 4 2 0 20

30

50

40

Number of Cars

60

70

8 6 4 2 2

80

4

6

8

10

12

14

Number of Consumers

16

Fig. 5. Average Delay (DTN)

Fig. 5 shows the performance of average DTN delay that represents response delay in DTN cases. We observe that DADT has longer DTN delay than other two DTN forwarding strategies. It is because all the strategies only retransmit once. In DADT, it waits for an appropriate time to retransmit Interest. Other strategies, however, retransmit Interest without judgment. Unlike average delay with instant connection, DTN response delay is significantly affected by number of cars but not by number of consumers. The reason is that in denser topologies there are more opportunities to find an appropriate neighbor to forward Interest. Also, the chance of hitting cache has little relation to DTN delay in our scenario. 10

Satisfication Ratio

Normalized Transmission Overhead

9

40 cars 50 cars

18

8

16

7

14 12

6

10

5

8

0.5

4 0.4 0.0

0.5

1.0

1.5

2.0

theta

2.5

3.0

3.5

3 0.0

30 cars 40 cars 50 cars

20

0.7

0.6

Average DTN Delay

22

30 cars 40 cars 50 cars

30 cars 0.8

6

0.5

1.0

1.5

2.0

theta

curve is more gentle than the others. It is because nodes will have less chance to find the next forwarder in sparser topology and the scope of the sector becomes relatively less important.

2.5

3.0

3.5

4 0.0

0.5

1.0

1.5

2.0

theta

2.5

3.0

3.5

Fig. 6. Impact of 𝜃

At last, we study the impact of 𝜃 on the performance. 𝜃 is used in equation (3) to represent the size of sector area to calculate neighbor spatial priority. During those evaluations, the number of consumers was maintaining 6 and we adjusted the value of 𝜃 to measure the performance of DADT under different topology densities. Fig. 6 shows the result of satisfaction ratio, normalized transmission overhead and average DTN delay. We observe that smaller 𝜃 will introduce higher satisfaction ratio, but lead to more overhead and longer average DTN delay. It is a trade-off that smaller scope of the sector will make the next forwarder selection more accurate, but lead to less opportunities to find an appropriate next forwarder, so it has longer DTN delay and more transmission overhead. Actually, when 𝜃 = 𝜋, DADT will have the same performance as First Contact, in which nodes pick up their fist encountered neighbor as the next forwarder. Therefore, selecting the value of 𝜃 can be a trade-off between achieving higher satisfaction ratio and gaining shorter response delay. Our another finding is that the value of 𝜃 seems to be less influential in sparser topology by observing that the green

In this paper, we have proposed DADT: a Density-Aware Delay-Tolerant Interest forwarding strategy to retrieve traffic data in vehicular NDN with the purpose of improving packet delivery ratio. It considers rebroadcast and retransmission together, also specifically addressing data retrieval during network disruptions using DTN. We have completed implemented DADT through simulation and demonstrated its performance by comparing against other DTN strategies. The results show that DADT achieves higher satisfaction ratio than other strategies without introducing much transmission overhead. We have also explored the impact of 𝜃 on network performance and found that choosing the value of 𝜃 can be a trade-off between achieving higher satisfaction ratio and gaining shorter response delay. R EFERENCES [1] A. Afanasyev, I. Moiseenko, and L. Zhang. ndnSIM: NDN simulator for NS-3. Technical Report NDN-0005, NDN, October 2012. [2] M. Amadeo, C. Campolo, and A. Molinaro. Forwarding strategies in named data wireless ad hoc networks: Design and evaluation. Journal of Network and Computer Applications, 50:148 – 158, 2015. [3] G. Grassi, D. Pesavento, G. Pau, R. Vuyyuru, R. Wakikawa, and L. Zhang. Vanet via named data networking. In Computer Communications Workshops (INFOCOM WKSHPS), 2014 IEEE Conference on, pages 410–415, April 2014. [4] G. Grassi, D. Pesavento, G. Pau, L. Zhang, and S. Fdida. Navigo: Interest forwarding by geolocations in vehicular named data networking. In World of Wireless, Mobile and Multimedia Networks (WoWMoM), 2015 IEEE 16th International Symposium on a, pages 1–10, June 2015. [5] V. Jacobson, D. K. Smetters, J. D. Thornton, M. F. Plass, N. H. Briggs, and R. L. Braynard. Networking named content. In Proceedings of the 5th International Conference on Emerging Networking Experiments and Technologies, CoNEXT ’09, pages 1–12, New York, NY, USA, 2009. ACM. [6] D. Krajzewicz, J. Erdmann, M. Behrisch, and L. Bieker. Recent development and applications of SUMO - Simulation of Urban MObility. International Journal On Advances in Systems and Measurements, 5(3&4):128–138, December 2012. [7] Y. Lu, X. Li, Y.-T. Yu, and M. Gerla. Information-centric delay-tolerant mobile ad-hoc networks. In Computer Communications Workshops (INFOCOM WKSHPS), 2014 IEEE Conference on, pages 428–433, April 2014. [8] M. Meisel, V. Pappas, and L. Zhang. Ad hoc networking via named data. In Proceedings of the Fifth ACM International Workshop on Mobility in the Evolving Internet Architecture, MobiArch ’10, pages 3–8, New York, NY, USA, 2010. ACM. [9] L. Wang, A. Afanasyev, R. Kuntz, R. Vuyyuru, R. Wakikawa, and L. Zhang. Rapid traffic information dissemination using named data. In Proceedings of the 1st ACM Workshop on Emerging Name-Oriented Mobile Networking Design - Architecture, Algorithms, and Applications, NoM ’12, pages 7–12, New York, NY, USA, 2012. ACM. [10] Y.-T. Yu, R. Dilmaghani, S. Calo, M. Sanadidi, and M. Gerla. Interest propagation in named data manets. In Computing, Networking and Communications (ICNC), 2013 International Conference on, pages 1118–1122, Jan 2013. [11] Y.-T. Yu, Y. Li, X. Ma, W. Shang, M. Sanadidi, and M. Gerla. Scalable opportunistic vanet content routing with encounter information. In Network Protocols (ICNP), 2013 21st IEEE International Conference on, pages 1–6, Oct 2013. [12] L. Zhang, A. Afanasyev, J. Burke, V. Jacobson, k. claffy, P. Crowley, C. Papadopoulos, L. Wang, and B. Zhang. Named data networking. SIGCOMM Comput. Commun. Rev., 44(3):66–73, July 2014.