Data dissemination in service discovery for vehicular ad hoc networks: a survey

: One of the challenges posed by the study of vehicular ad hoc networks (VANETs) is the transmission of data issued by valued traffic information services under incomplete link conditions. Many dissemination protocols have been developed by the community to solve the issue. In this survey, the authors explore the service discovery where data dissemination should serve as a foundation. Then, they propose an overview and taxonomy of a large range of data dissemination available for VANETs. Finally, they illustrate the simulation infrastructure by collaborating two independent simulators. The objective is to provide guidelines to easily understand and extend the capabilities of protocols according to the users’ needs.


Introduction
In opportunistic networks, it is not necessary to link sources with destinations using complete channels.Data transmission relies on the opportunities that occur when the nodes are connected.With other traditional networks, the best-known quality of an opportunistic network is robustness.For example, a method of routing is expected to adapt constantly to the changing circumstances.In addition, it facilitates data dissemination under conditions in which the links are incomplete and the predefined forwarding nodes are not available.
Vehicular ad hoc networks (VANETs) use the basic principles of opportunistic networks, enabling moving vehicles to reroute from traffic congestion.Through such context-aware and tolerated dissemination, the vehicles can carry and send data packages to roadside infrastructures or other vehicles.However, it is difficult for routing technology in mobile ad hoc networks (MANET) to accommodate traffic environments.First, defining the sources and destinations of data packages in advance, as required by MANET, is practically impossible.Forwarding nodes for routing are set according to current environmental attributes, including rapid traffic-flow changes, fixed or regular vehicle trajectories, and fewer delays in dissemination.Fixed infrastructures are then prompted to be attached with a data-processing system and to provide a reliable power supply.It enables these data packages to be stored, analysed locally, and released when the appropriate vehicles approach.Finally, it is necessary for VANETs to impose stringent durability and availability requirements; as such, these technologies should work well under unstable network situations to ensure smooth delivery of these data from end to end.
One practical application of data dissemination has been in service discovery for VANETs, where information about traffic services is collected, appraised, stored, and updated.After a few years of exciting developments, various technologies have been applied to data dissemination, ranging from the most trivial to the most realistic.However, as each research group basically develops its own specific needs, these developments lack coordination.For example, such developments include accommodating fast-changing data flow, or promoting broadcast quality, or adjusting transmission protocol to facilitate communication among vehicles, storing information, and managing data.
In this paper, we provide a detailed survey and comparison of the dissemination technologies available for VANET.First, in Section 2, we introduce a framework of service discovery where data dissemination should serve as the foundation.In Section 3, we detail the criteria used to classify the various methods.In Section 4, we propose a taxonomy of methods for routing.A brief tutorial on how changes to travel, vehicle, and network can contribute to data dissemination is presented in Section 5. Section 6 describes the performance evaluation of simulation methods, and Section 7 concludes the survey.

Service discovery in VANETs
In VANET, real-time traffic and traveller information services promise unimaginable levels of personal mobility using road status and alarm information.It also poses unique service discovery challenges [1].For instance, a traffic setting might be saturated with hundreds of vehicles.The roadside devices provide information to users without their active attention.To contribute to the changing needs of travellers whenever necessary, regional and architectural differences must be taken into consideration in rationally allocating and using the technical resources from traffic information service (TIS) providers.Thus, service discovery is essential for achieving such a level of sophistication.
Although many organisations have designed and developed discovery protocols over the past few years, these protocols support service discovery for home or enterprise environments and thus do not always apply to more dynamic and heterogeneous traffic environments.Service discovery in VANET differs significantly from web service discovery over the Internet, which has no physical location limitations.Nitto et al. [2] demonstrated major service discovery protocol components, of which the 'communication method' fits within our scope.It seemed impossible to communicate in VANET by unicast, multicast, or broadcast only because the network is unconnected in most cases, resulting in the loss of data delivery.An alternative method working with different kinds of information in traffic is to establish a temporary, single-hop, and local routing from a source to one or more destinations.Then, all data from the source station are transmitted hop by hop to the destination(s) several times over the duration of the local route.

Data dissemination
One aspect of data transmission for service discovery is data dissemination.Traditional data-disseminating mechanisms cannot work well for the following reasons.First, the moving patterns (e.g.speed, direction, and the stages of rest) of many vehicles in VANET can affect the quality of the data dissemination.Second, high travelling speeds can alter traffic-flow density and affect the transfer of data.Finally, when the communicating equipment is not running or the road obstructions cause traffic to divert, the signal becomes attenuated.In consequence, traditional communication patterns among fixed routers become invalid due to the special structure of VANET.
VANET has three main types of network structures.(i) Under wireless wide-area network (WWAN) cellular mode, the fixed communication equipment, such as the base stations, are set up around the whole region and form a specific topology.With this type of access point linking, each moving vehicle communicates through the stations nearby and delivers the information to switching centres through the cable network.Information can be delivered to the wireless stations for storing and disseminating sometime in the future.(ii) Under the ad hoc mode, the wireless links emerge among the vehicles only, with no predetermined strategies about how to select the next hop to receive and forward the information.Instead, information delivery occurs only when vehicles meet, and vehicles can transfer, secure, and forward the data.(iii) Under the hybrid mode, both the vehicles and the fixed equipment easily intercommunicate.Fixed equipment might include sensors, base stations, routers, and roadside and mid-road units.These components comprise the topology of the VANET, in which the fixed equipment provides the foundation and the moving vehicles forward the information and compensate for any lapses in the fixed equipment.Table 1 summarises this comparison of the above three types of network structures for VANET, where the roadside unit (RSU) includes fixed devices along roads, such as base stations and sensors.
After receiving the demand for the TISs, the location of the provider serving the goals and requirements is still unclear.Thus, the locations must be determined in time to disseminate and convey the service data accurately [3,4].In our early work, we classified data dissemination of how to address the needs of services for vehicles into two categories: data subscription and data retrieval [5].

Data subscription:
In VANET, vehicles subscribe to TISs, and the providers, that is, the vehicles or roadside devices, will announce to subscribers when the desired information is produced.This type of data dissemination works based on the subscriptions of the moving vehicles, which can be classified into clusters to allow plenty of replicas to be transmitted.Such replicas can be delivered to the clusters moving in the reverse direction, thus meeting the needs of a larger number of possible subscribers.In Zubi and Krunz's [6] study, the travellers can subscribe to the data of interest and provide descriptions by using a quadruple notation.Such a notification message with a chosen topic can be defined and include a disseminating region and an available time.When a vehicular node holds one piece of the data, it broadcasts its notification and waits for the responses from other neighbours.After gathering all the necessary information to create a publication, the replica holder determines the direction and number of the disseminated replica.A broadcast publishing model for larger-scale road networks is given in Zhang et al.'s study [7].The model is a part of rapid dissemination process where data collection, fusion, and display are studied.In view of the unreliable transmission, Zhang et al. found that vehicles always receive and send the congestion information repeatedly at a broadcast interval.

Data retrieval:
Data retrieval is driven by the TIS requester, which appears as a vehicle in VANET.Unlike data subscriptions, vehicular nodes post-detailed information requests on the network, and the nodes receiving the information search their memory and reply with the available data.There are two data-retrieval mechanisms, one in which the dissemination occurs with the help of intermediate equipment, and another that uses 'end-to-end' dissemination between two vehicular nodes.Wu et al. [8] describe a routing method called geographic load balancing routing in hybrid VANETs, which ensures greater reliability throughout the system.It consists of a mount of routers, base stations, and other roadside WI-FI access points, which are established as the significant area role (AR) for vehicle-to-roadside and roadside-toroadside communications.Each vehicle has a local AR for intercommunication.Requisitions from the source nodes have always been transmitted to the local ARs, which are responsible for continuous data collection.The local ARs transfer the data to the designated vehicular nodes with the help of other ARs.In Dua et al.'s [9] study, all the vehicles were clustered according to their storage and computation capacity.The information was retrieved from the nearest RSU first.Otherwise, a vehicle acting as a cluster head placed within the transmission range was expected to inform the data retrieval.The scheme for retrieval was found to work well by using only (wherever possible) the vehicles with a high processing capability.

Taxonomy: description of the criteria
For a detailed description of data dissemination during the discovery of TIS, it is necessary to define the key criteria.In this section, we introduce a set of criteria extracted from the framework described in Fig. 1.The proposed criteria fall into three major categories: people, vehicles, and roads.Obviously, the more criteria a method includes, the more efficient it is, and the more realistic the situations it supports.
(1) People: It is practical for TIS to focus more on individuals whose primary concern is travel.It is possible to use a filter for personal purposes to reduce large amounts of traffic contextual data and to make the TISs the most easily accessed.In particular, the personal behaviour is monitored and saved to serve as a reference for data dissemination among the vehicular networks [10].Factors such as the environment, living conditions, and social relationships are all taken into account when devising a propagation model.A still more advanced technique, called social awareness, actively creates solutions for perceiving and recognising individual interests and analysing characteristics such as mobility patterns in VANET [11].Such significant patterns tend to be clear and reasonably stable, which reduces the time and effort needed for analysis [12].In Fig. 1, the recognised purposes indicate the preferred destinations where data should be transmitted.Then, special propagations are identified.For instance, an emergency has to be disseminated to those vehicles affected in all directions of a road network using the broadcast.In this case, the destinations are not defined in advance.Once the TIS providers (destinations, e.g. the stations that have the fuelling services for 76 octane petrol) are given, the data from the TIS requester should be transmitted hopby-hop with the help of the intermediate equipment.One of the two propagation models should be identified so as to make data dissemination more efficient.
(2) Vehicles: Vehicular nodes act chiefly as the producers and receivers in VANET.In addition, they serve as a means of carrying and forwarding information.Vehicles tend to move and interconnect, which leads to inhomogeneous distributions of traffic movements.Therefore, knowing the key characteristics of the vehicular nodes and drawing their motions is necessary to identify direct or indirect routing schemes in VANET, as shown in Fig. 1.Vehicular motion can be classified into three categories.The first one is the movement of purpose-built vehicles, such as a bus.Buses share fixed routes and timetables; thus, it is easy to identify mobility patterns according to their regular routines.The second pattern is concerned with the movement to a given destination, presumably via taxis.Optimal routes can be found between current locations and destinations with the help of electric maps.As indicated in Fig. 1, such mobility patterns illustrate vehicles moving through the shortest paths skewed by personal preferences.The last category is the motion guided and altered by the road network [12].Vehicles move to their destinations, which are selected at random, and often through indirect routes, for example, to avoid a town centre.It illustrates probable directions for different routes [13].These motions should be considered when designing opportunistic routing.
(3) Road: The main difference between VANET and MANET is that the vehicles move along roads that were constructed in advance and are the most stable of the above-mentioned categories.Thus, it is important to establish the road network as an indicator of the direction in which the data are travelling, then the vehicles can serve as carriers for delivering data according to the topological maps and to the RSUs promptly.Destinations such as coffee shops, book stores, and so on are assumed to be stationary [14].Otherwise, data are passed and cached among RSUs, such as sensors and data processing centres, until needed.Thus, these units can be used as the data relays and can facilitate the opportunistic routing.In terms of geographic knowledge, topological maps are the primary concern within numerous routing strategies.In particular, it is necessary to observe the geographic atmosphere around the vehicles, for instance, either a vehicle locates an intersection or general sections, and to define the probability at which a vehicle would travel along a given road section.Furthermore, considerations about the suitability and availability of a road are also necessary.As illustrated in Fig. 1, the length of each section has played an important role in assessing the traffic capability [13].The greater the traffic density is along a section of road, the less traffic capability it will provide.Thus, another important consideration is the width of each section, because it can be used to determine traffic densities, vehicular mobility models, and velocities of data dissemination.For example, any difference between the narrowest and the widest sections could affect data delivery [14].

Routing schemes for vehicular networks
The proposed guidelines illustrate some of the most realistic configurations.Most protocols refer to the vehicular mobility trace or, at least, road conditions as routing constraints.Routing, opportunistic forwarding in particular, remains of primary significance in data dissemination.Here we describe the challenges specific to routing in VANET and provide examples of solutions related to the number of hops and candidate forwarders.

Number of hops in routing
It is necessary to devise routing with single-hop or multi-hops in mind.One method defines the layer at which the forwarding occurs.Some routing protocols demonstrate how data would transfer among neighbours, whereas others deliver data every time by selecting the next nodes that might be several hops away.Liu et al. [15] applied the idea of multi-hop routing.When producing data, a source vehicular node can obtain road conditions and all vehicular mobility traces in VANET.Then, the source proceeds along the most economical path to its destination, keeping the nexthop nodes forwarding the packet to a minimum.Thus, such nodes serve as a foundation to direct a path for data dissemination in VANET.When a vehicle diverts from the expected path, nearby nodes will be chosen.To reduce the number of hops, forwarding in groups is effective.In Zubi and Krunz's study [6], each forwarding is made within two hops, that is, the packets remain active during the two-hop transfer.To build a routing table that contains all onehop and two-hops items, the source broadcasts the message 'HELLO' and gives its location and ID.The one-hop nodes transmit their information and decrease the time to live (TTL) to 1 after a packet arrives.Meanwhile, the packets continue to deliver towards the two-hop nodes away from the source.Now, TTL reaches 0, and this node transmits a reply to the source.The table is then used as the source to assess the status of the two-hop route.
Most routing protocols for opportunistic VANETs tend to regard data transfer within one hop, in which one-hop forwarding should be settled.Musolesi and Mascolo [16] assessed all the possible alternatives of each neighbour to choose from and enables hop-byhop relay.A forwarding table is then set and exchanged periodically for facilitating the update, and each node can disseminate data to its one-hop neighbours chosen from the table to accomplish a single transmit.In topology and link quality-aware geographical opportunistic routing protocol (TLG) [17], the vehicular nodes broadcast data, which define the geographic distance from their candidates to the destination, and compare these distances with those from their last-hop neighbours to the destination.If the former is shorter, these candidates will carry the data until an available node is determined in the competition, that is, the other nodes set the timers and calculate their dynamic forwarding delay (DFD) based on the link status.The earlier their timers hit 0, the greater the probability the nodes will become the next forwarder.Thus, all the forwarders can be selected in the same way, and the request can be disseminated along the paths consisting of all forwarders.

Opportunistic forwarding in routing
Extensive knowledge of the network is essential because the network serves as an excellent bed for TIS service discovery.Naturally, knowledge of these networks influences the selection of a range of available vehicular nodes for collaborative transmission.The intellectual interests concerning this topic encompass the selection of next-hop node(s).The scheme of opportunistic routing in VANET is store-carry-forward.That is, a vehicle stores data packages by the order received and carry them onboard until next transfer is determined.Thus, the selection process should be handled at the proper time and in an appropriate manner.Numerous attempts have been made to promote the quality of data dissemination.Globally, the decisive factor of the selection algorithms might be classified into three different classes, which are illustrated in Fig. 2.

Fine-grained (node) selection algorithm:
Major studies have focused on selecting the available vehicular nodes affected by the location of the nodes.Al-Zubi and Krunz [6] treat the source as a decision point and issue a transfer along the route with a minimum of forwarding nodes.The source discovers a route with three hops, that is, the data package can be sent successfully across a logical channel (including two forwarding nodes) and, finally, reach a destination.These nodes consist of the skeleton for one transmission.Actually, because of vehicular movements, data packages are likely to be delivered to the other nodes, which issue the beacon frames to announce receipt of the packages around the skeleton.Therefore, information will be delivered with hop-by-hop forwarding among the pre-set vehicles identified by the source.
The intermediate nodes included in a path that connects the source and destination can also lead to a selection.In Dau and Labiod's study [18], the source is required to install the identified forwarding trajectory into the packet header and launch the packets to its one-hop neighbours.A selection scheme enables neighbours to compete for a transmission using their timers.This process can be gauged by comparing the geographic distances from the trajectory and from the last forwarder, respectively.Obviously, when a node has the least distance from the predefined trajectory and the longest distance to the last one, as well the advantage of a good channel, it can set the quickest timer and broadcast the packets.Wu et al. [19] aimed to make full use of the reception diversity to improve the quality of data dissemination.The source first defines a region to decrease the numbers of replicas, and then the receiver, which stays farthest away from the source, gains the control to transfer as the next forwarder.This method also has the benefit of reliability.In general, the intermediate vehicles nearest to the source decide whether to deliver the packet, and then the requisition can be transmitted hop-by-hop in the same pattern until it reaches the destination.Chang and Lee [20] proposed a distancebased routing protocol for intersections.Three prioritised zones were identified for locating the potential forwarding nodes, and these zones took into consideration the vehicle's position and intersection waiting time.Thus, the nodes have different abilities for broadcasting a packet.This approach allows the stable node, which is expected to have, wherever possible, the same relative velocity as the sender, to be selected for the routing path during the packet traversal time.

Coarse-grained (road) selection algorithm:
Transmission of a data package along the roads in VANET is special because the roads are laid out so regularly.Road and associated geographic information is then expected as the decisive factor.Generally, the road topology can serve to direct the packets towards the roadside.Leontiadis and Mascolo [21] presented a scenario where the navigation system helps to select the next carriers.A routing is expected to consist of the vehicles that can travel the shortest distance to arrive at the given destination.Therefore, the minimum estimated time of delivery (METD) is estimated assuming each predefined node as the next forwarder.Finally, the vehicles that minimise the METD can be selected; these control the transmission in the next round.In this way, the decision to issue one transmission depends on the road topology and vehicular movements.
The intersections that are available for data transfer are another kind of road topology.Designers can identify more opportunities for forwarding because of the more information available at the intersections.Road segment conditions can be estimated using the delivery delay from each vehicle and be updated quickly with the most recent estimation according to time [14].The method by which the information can be transmitted consists of three definite, practical phases, viz expected routing, packet buffering in the road, and exchanging at the intersections.In more specific terms, the vehicle might choose the most ideal route and propagate the packets towards the back-and-forth neighbours to provide more opportunities for switching.Two options are there for duplicate propagation on those who move along a road.One is related to the vehicles moving along the direction of the route; the other is related to those moving in the opposite direction.In either case, by increasing the number of duplicates carried by the vehicular nodes, the chances of success in multi-hop data delivery increases.

Mixed-grained selection algorithm:
Data dissemination exhibits characteristics of several different identities, i.e. vehicular nodes and infrastructures, in the hybrid vehicular network.Opportunistic routing facilitates communicating among the distributed vehicular nodes by tracking where they go and when they forward.Much of the infrastructure that serves as the backbone of VANET is used for the data relay.Thus, good design of data dissemination can either deliver the packet from one vehicle node to another or can upload to the nearby infrastructure [22].
Wu et al. [8] spread the network load over the routers and vehicles.A protocol is given for switching three routing modes naturally, depending on situational changes in context, e.g.location and network connectivity.As the backbones of VANET, wireless ARs are fixed and contribute towards better traffic load and more reliable transmission.The connections involved between a vehicle and an AR have been made by delivering data as close to the destination as possible.Two plans are then presented for treating AR and the vehicle as the destination, respectively.Another kind of connection that works well for a hybrid VANET lacking available ARs, enables the transfer of data among the expected next-hop carriers.Napit and Trivedi [23] classify the equipment into an onboard unit (OBU) and a RSU.Also, the three schemes can facilitate the communications between such RSUs.The data will be delivered to a base station nearby from a carrier and then forwarded to another vehicle within the range of the base station.
Considering the highly variable topology of a vehicular network, numerous activated sensors can function in the low-density scenarios.It is the nearest sensor's turn to disseminate the data and to arrange a plan for forwarding.In this way, multiple devices falling into OBU and RSU categories will better function with one another to enable data dissemination.

Discussion
In this section, we illustrate the available approaches for routing in VANET.Given the complexities of traffic environments, various methods have focused on the real-time geographical location of vehicles.These approaches are reliable and work well in hightraffic density for end-to-end data dissemination.Vehicular nodes might predict the chances of accessibility to a destination and make appropriate relay decisions.For example, least-length path routing (LLPR) integrates the direction and a minimum distance from a line connecting the source and the destination [24].For smooth traffic regions, one solution would be to approximate the chances with road topology, e.g.width of road segments [25].In GeoSVR, a rectangle is split to cover the route, the property of which is affected by road capacity.This approach is limited by the modelling of global road conditions, owing to the auxiliary prerequisite dependence.The routing also lacks effectiveness without such a prerequisite that is not observed by the actual traffic environment.However, this method has the advantage of being able to establish stateless routing under predictive road conditions.We provide in this section a brief description of the major routing strategies available to the vehicular networking community, which are presented in Table 2.However, both approaches are not mutually exclusive, as the road topology depends highly on the flow of traffic.For example, the urban map alone will not give a satisfactory transmission; realtime position must be used with it.Conversely, routing makes use of exactly the type of beacon, usually the positions of vehicles and infrastructures, that might easily encounter the 'local maximum' problem [27].In fact, it is necessary to carry out many ways of organising and structuring roadside infrastructures for data dissemination, according to the applications concerned, e.g.fire accidents and road crashes.

Identifying data dissemination for TIS
Data dissemination is one-to-all communication in a special region, whereas routing is one-to-one communication towards a certain direction [28].Panichpapiboon and Pattara-Atikom [29] had reported some findings on data dissemination.They believe that disseminating information relies on broadcasting protocols [30], which they classify as single-hop and multi-hop modes.Daraghmi et al. [28] explored the solutions for serving well-known communications in single-road segments or between different road segments.We provide a brief tutorial on how the changes in travellers, vehicles, and networks contribute to data dissemination, depending on the criteria for the taxonomy introduced in Section 3. The major dissemination protocols are classified in Tables 3 and 4.

Dissemination based on travellers' needs
We could try to channel the data in a direction that is advantageous to diverse travellers for TIS.In view of the concerns expressed in the introduction and the need to disseminate more effectively, it is necessary to mine the details about what travellers are looking for, and transfer that data on purpose.Two main methods serve to gather the travellers' needs, one is to locate the behaviour patterns from the routine, and one is to accumulate the subscriptions in advance.For example, Leontiadis and Mascolo [10] attempt to issue a dissemination to all interested receivers in an area.A navigation system is then designed to offer a subscription representing what travellers desire, which indicates the area in which the data need to be disseminated.Accordingly, the dissemination might happen making full use of replica forwarding between vehicles.Moreover, the vehicles will carry and forward until there is at least one interested vehicle in another region, preventing unnecessary data replicas.In Pandey et al.'s study [43], some stations that acted as the publisher, subscriber, or broker from area to area, were installed at major intersections to facilitate the dissemination.Similarly, subscriptions are generated to indicate travellers' interest in receiving certain types of data and can be paired with the brokers using the distributed hash table.Each vehicle continues matching and hop-to-hop forwarding the clustering vehicles in the direction of travel; forwarding according to the subscriptions from any vehicle in each cluster random routing, epidemic routing replicas required, delivery ratio BAS [14] determining the direction in which the neighbours travel; data will be transmitted in two directions GPSR, epidemic routing delivery ratio, delivery delay road segments GeOpps [21] forwarding to the node with the shortest distance from the destination MoVe, Greedy delivery ratio, average packet delay, average hops for different network density LLPR [24] identifying the next-hop that has the angle subtended between the motion vector and the anchored node vector MoVe end to end delay, transmission distance road-conditionbased forwarding width of road segments GeoSVR [25] identifying the optimal route by computing the average width and deviation of each segment; selecting the next-hop node along this route AODV, GPSR delivery ratio, hops decision-makerbased forwarding current node OOMM [6] determining the skeleton with Dijkstra by the source; selecting the next-hop node along this path FBSA, RBSA [26] With the publish/subscribe paradigm, data dissemination could work well through subscriptions from each moving vehicle because of a reduced demand for delivering unnecessary data and therefore providing an available pool of bandwidth.Generally, vehicular nodes will be classified into clusters, and replicas will be transferred among the clusters, especially travelling in the opposite direction to locate more interested subscribers.Measuring, describing, analysing, interpreting, and transmitting such useful data can be helpful.However, only one cluster can receive the forwarded subscriptions, which limits efficiency in the transfer if other clusters are interested.In addition, because of the dissemination mode resulting from the accuracy of topics, another challenge is how to describe a subscription.Topic is usually related to the location of interested points and the key items of applications.A full and accurate picture of topics is useful for locating the service offers and directing a path for dissemination.However, putting this principle into practice might require more prior knowledge, which is difficult because of the uncertain contexts in VANET.

Dissemination based on vehicles' movement
VANET differs from traditional networks because it can observe and mine the mobility patterns of vehicles travelling on urban streets.Thus, information on how a vehicle moves is requested to make sure that data will be transmitted efficiently and correctly [44].For instance, the purpose of a navigation system is to find the facts in a particular situation, and therefore a solution, so that the work of locating a vehicle and suggesting paths should progress in the most effective manner possible [45].The time to deliver the reply would utilise the vehicles' movements.Traditional dissemination protocols cannot work well due to uncertain network topology, where the replies are issued and need to be transferred as effectively as possible.To create an effective dissemination facility, Leontiadis et al. satisfy as many of the mobility modes as they can expect by using the navigation system.The info-station collects the service requests and further delivers them to the vehicular node, which will follow the predefined motion.The protocol keeps the reply on the vehicle travelling towards another vehicle along the predefined path, which will increase the encounter probabilities.The weakness of the protocol in predicting the movement of vehicles is clear, particularly when the motions and connections are so changeable.Also, the protocol has not accounted for the fact that there is probably no vehicle that can serve as the next-hop  forwarder at the time a vehicular node moves along the predefined path for carrying the requested data.Therefore, an ideal prediction at the expense of constantly changing contexts in VANET is problematic.In an increasingly complicated traffic world, more vehicles can travel in an undefined manner than not, which challenges such disseminations because they are set up to operate with the expectation of advanced knowledge of vehicular movement.However, by classifying the movements observed [12], buses have unique advantages as information carriers.It is difficult to extract mobility pattern from vehicles that require large-scale multidimensional and time-series traffic data, thus, the primary focus is exactly on how buses would disseminate data in VANET.Ma et al. [46] use the buses as transmitters, and each one has a set of targeted road segments (TRS).Once the service requests are reached, the buses will refer to their existing TRS list immediately and deliver the data packages to the other buses if the desired destination is not included in their own TRS.Furthermore, it configures forwarding mechanisms by selecting a bus that will move towards the specific destination with the 'closest distance predictability'.Unlike most existing dissemination protocols, the author predicts the angle between the motion and distance vectors to determine the candidate carrier.Fig. 3 illustrates how these vectors work and the angle of the two vectors can be calculated as (1).The distance along from the current location B and destination D is calculated based on (2).Thus, the bus whose predicted value is the shortest will sever as the disseminator.The downside to this approach is that the way of dissemination developed is based on the mobility pattern of buses, which limits the design to simple cases.Also, obtaining knowledge of which bus can be selected as a candidate, without the use of movement time of buses, is no guarantee of efficiency in delivery because of the differences in the travel speed, although this bus is nearest to the destination Some solutions provide opportunistic forwarding by classifying vehicular nodes into clusters with different characteristics.In this way, data packages would transfer from one cluster to another [34].Momeni et al. tried to group clusters using the 'HELLO' message transmitted among connected vehicles for a predefined period.The request would disseminate from one cluster header to another through the routing tables to find the candidate forwarders moving in the same direction towards the destination.A vehicle's movement described with its location and direction is listed in a routing table.For cluster-head choosing in VANET, Tian et al. [47] adopted the distance between two vehicles to dynamic partition different clusters and used a gradient optimisation process to find the suitable cluster head.Delivering a significant amount of data among a few cluster heads and gateways is problematic because the cluster heads are considered fixed or unalterable.The established cluster heads could not communicate with other vehicular nodes if a link experiences short-term congestion, and the nodes will not acknowledge the receipt of control messages from newly elected cluster heads.Over time, this condition could increase the number of clusters that would threaten classified administration.Leontiadis and Mascolo [10] define the clusters using the vehicles' movement, especially the directions towards which the vehicles gravitate.Fig. 4 gives a diagrammatic representation where the vehicular nodes on road can be grouped into four clusters, which are marked with different symbols.The direction in which a vehicle moves determines the cluster of which it might be a member.Subscriptions can be disseminated among clusters by using the carriers that oversee selecting clusters to forward, according to the number of nodes that are uninformed and interested in a specific cluster.When the replicas from publishers are ready, they can be delivered directly to the cluster, which has maximum number of subscribers, and the vehicle moving towards the opposite direction can be located at the next carrier.Thus, the results of data dissemination can improve productivity.
Unlike existing works that consider mainly mobile vehicles, efficient data dissemination protocol (EDP) leverages parked vehicles along roadsides for disseminating data.EDP groups the parked vehicles into a cluster, which buffers and relays data from mobile vehicles [48].Although introducing some vehicles as roadside infrastructure was worthwhile, dissemination could be interrupted when the parked vehicles begin moving.

Dissemination based on urban street networks
Designers have always considered describing the random motion of vehicles to be a much more difficult.Instead, they have searched for alternatives as to how the data can contribute to VANET.Providing rich and in-depth information, mostly geo-referenced data relating to street networks, are then thought of as a predictable identifying mark when forwarding a service-request packet.Due to the end-to-end dissemination, data such as the geography are so reliable that they apply well in the context of high-traffic density for delivering emergency service requests.In [24], vehicles might predict the location-based probabilities of the candidates reaching their targets for transferring, and the best one can be selected.Namritha and Karuppanan defined the shortest path as the sourcedestination line and used GPS to provide the real-time position.LLPR is then proposed to disseminate opportunistically where the candidate not only has moved towards the fixed node (AN), indicating where the accident happened, but also has the minimum distance to the line.The former is recognised in the presented approach/depart algorithm, and the latter determines the forwarding vehicles in developing the near/far algorithm; both are illustrated in Figs.5a and b, respectively.
The instantaneous directions in which the vehicles are travelling are measured by in-car compasses and provide the motion vectors for the selection of potential forwarders.In Fig. 5a, the angle between the two vectors is calculated to assess the probability that a vehicle is moving towards the AN.To identify the candidates, it is necessary to capture the deviation from the shortest path shown in Fig. 5b and to select the one with minimal deviation.This method relies on the idea that the potential forwarders must stand on the streets with the direction of travel now corresponding to the direction of AN.However, if a candidate were to meet a substantial obstruction in passing across the roads, it could quite easily deflect; at worst, it could turn around, which could make the selection of forwarding ineffective.
Another idea is presented in Leontiadis and Mascolo's study [21].By designing the dissemination for different TISs with different specific goals, we could create designs that deliver source-destination data based on the geographic information.The same holds true for the design of transferring data back.An in-car navigation system can recommend the routes regarding the position of destination provided by the neighbour, the driving preferences, as well as the road conditions.In Fig. 6, when a carrier moves to P 1 and following P 2 , it meets three moving vehicles.The navigation system of each one calculates the most likely paths, marked a, b, and c, respectively.It is obvious that three vehicles will move separating in every direction.The points shown as NP a , NP b , and NP c , directed towards destination D on each route with the minimal estimated time of arrival, are determined.As illustrated in Fig. 6, the carrier selects the vehicle travelling along path c as the next forwarding node.GeOpps only applies to the selection of the next-hop carriers, however, no provision about how the replication is established.Therefore, only one carrier will be chosen in every estimation, which involves significant data loss because of bad connections.
Urban street networks usually come with an assortment of city blocks having different road topologies.The manner of data dissemination in VANET forces it to apply the road topology.GeoSVR [15] regards the width of the specific road segment as a predominant influence in the numbers of vehicles passing through and, therefore, in the delivery of service-request packets.A route with the most traffic is the best candidate for dissemination.Specifically, the source will define a rectangle located between itself and the destination, which lays out fields for data delivery.Each carrier might assess the road capacity to pick out the best route by using the width of all road segments and the deviation of each segment.As shown in Fig. 7, a rectangle including two paths is marked as AB-BC and AD-DC.Two paths are represented as (A, B, 60), (B, C, 60), (A, D, 20), and (D, C, 100).It can be determined that AB-BC with the minimal deviation is finally selected as the best route, although the widths of all road segments belonging to AB-BC and AD-DC, respectively, are the same.However, GeoSVR lays one-sided stress on the road width in decisions, without accounting for the length of a segment.Moreover, there is nothing in the elimination about identifying the next carrier and the number of replications in GeoSVR entirely, which may result in losing sight of the congestion conditions caused by the data transfer on a predefined route.The local road network and rapid dissemination (LNFRN) [7] protocol allows different types of data to be disseminated, depending on the road sections.The original information, such as the distance travelled, is published to other vehicles in the same section.Vehicles broadcast the simplified congestion information fused to the vehicles in other sections.With the help of link nodes, LNFRN enables the congestion information to rapidly cover the entire urban road network.Yet, LNFRN is not applicable to the case where vehicles are located on a cross street.It also requires that the information about road sections is known in advance, which limits its practical use.

Dissemination based on link state
There are other constraints such as bandwidth, velocity of transmission, and link quality on the VANET, which restrict the freedom of data disseminations.It is possible to account for relay in vehicular opportunistic networks, which is affected by link status information.The idea in TLG [17] is to design a protocol that integrates the geography and links to form a cohesive dissemination process.First, a vehicular node compares the distance at which it travels under the influence of traffic movement to the destination, with that of the last forwarder.If it is the one of two that is less distant in space, the node will act as the candidate and calculate its own DFD [17].The indications, i.e. remaining energy, link quality, and projection progress on the sourcedestination line are introduced for measuring the delay.The node that holds the least DFD can precede the disseminations, whereas the other nodes will drop the packages.Zhao et al. [49] developed the same idea of a link-aware dissemination.Four link-related indications, i.e. quality, progress, energy, and validity, are proposed to assess each DFD.What differs from the method mentioned previously is the computation of these indexes.A consistent matrix is used to define whether one index is important.For instance, '1' indicates both are of similar importance.Due to the mercurial changes of links in VANET, context-aware adaptive opportunistic routing (CAOR) applies the analytic hierarchy process to weigh the same index differently, depending on the motivations for the advanced knowledge and the ITS applications.
In assessing each wireless link in VANET and determining the measure to be applied for achieving the best route, the relevant factors -available bandwidth, link quality, and vehicle movement -are taken into account in the 'next-hop' selection in portable fuzzy constraint Q-learning protocol established on the basis of AODV (PFQ-AODV) [50].Neighbour information is especially needed for understanding vehicle movement because the traffic environment is fraught with uncertainty; furthermore, a vehicle's position is not always available.By identifying the logical connections between each set of information from the 'HELLO' messages, it is possible to infer the link status that leads to the next forwarder.By taking advantage of fuzzy logic in providing flexibility and complexity solutions, PFQ-AODV creates the best route based on the quality of a candidate using Q-learning model.Like the original AODV, data will be delivered along the temp path traversed by the reply packet, which makes dissemination capable of adapting to the dynamic vehicular network.However, because of the lack of unified rules of link status, rules are formulated according as specific needs, resulting in different metrics and leaving room for discretion.Such link-aware dissemination is affected by many factors; thus, it is difficult to obtain satisfactory results in real-world traffic scenarios.

Dissemination based on contexts
Contexts describe any entity involved abstractly.In VANET, contexts could be any factor mentioned affecting data dissemination during service discovery.The central theme of letting contexts drive the dissemination is to adjust one or more contextual properties from time to time as the settings change.There is a need to consider what contextual operations and variables will best meet the needs of delivery and what the implications of these decisions might be on data dissemination.For example, CAOR makes a contextual decision with four different elements and identifies their respective weights in time.Of these elements, three describe the link conditions, and another one concerns geographic position.These contextual elements will, in practice, impinge on one another.To make the elements effective in routing, CAOR assigns the relationship of every two elements using a weight matrix and tailors the matrix according to the situation of the field.For the context-based transfer, different contextual information can be used to describe the living conditions and social relations of each user in the history-based routing protocol (HiBOp) [51], and be saved in an identity table (IT).The newly identified neighbours can be recognised by exchanging their own ITs on demand in a broadcast.Specifically, the source attaches the IT of destination with the request packets, and then nodes in a network disseminate the data by calculating the probability, that they reach the destination and meet a node carrying the desired information.This process can occur by using the contexts included in the IT and history tables, respectively.HiBOp solves two issues, that is, historical contexts can be learned for discovering the available contexts in a new setting, and decisions can be made based on good next carriers.It does, however, make work more cumbersome than necessary for the HiBOp because each encountered node should be saved and should match the IT of destination with its own IT.It then will cause a heavy load within the whole VANET if there have been many destinations providing remarkably diversified TISs.Also, it is problematic to control the replication in HiBOp, which could involve significant data loss.
Due to the changeable quality-of-service (QoS) metrics of routes, Hashem Eiza et al. [52] use the situation-aware (SA) model to provide a reliable data dissemination, which characterises a situation by obtaining contexts with different styles of topical subjects.Three stages in the SA model cover vehicle-related data recognition, link/routes prediction, and QoS guarantees.The ant colony system rules are applied to determine the optimal route for data dissemination, together with a developed rule, named as the 'QoS awareness' rule.Generally, several pheromones indicate the quality of the candidate routes, which also indicates whether this route towards the destination will disconnect.In this case, pheromones are subject to evaporation, and a backup route must be explored.Although the method has claimed that it can work well for multi-constrained QoS routing, plenty of routing tables should be managed in each vehicular node, which leads to an increased demand on computing and communicating for nodes.It does not fit the paradigm in VANET very well.
Due to the unique characteristics of VANET, this type of dissemination typically uses two or more specific contexts, e.g.people, vehicle, and road, and link-related.They improve the reliability [53,54], quickly relieve the interruptions [55], and continue the quest for good routes [56].However, as focus expands to encompass more and more situations, it becomes increasingly difficult to create a concise and coherent set of contexts that represents the diversity of real traffic environments.This situation can also be a challenge for data dissemination, given the contextual limitations on routing, delivering, and forwarding.

Discussion
Researchers striving to design the best TIS-related data dissemination programs have different focuses and bases.A proposed classification is presented in Table 3.
The different protocols are generated by using specific constraints and hops.The one-hop routing used most often is based on the locations of vehicles under road conditions during data dissemination, that is, the next carrier can be determined hop by hop.A few other protocols, e.g.topology-assist geo-opportunistic routing (TO-GO), determines the next two-hop carrier each time when one local transmission is terminated.The few ways in which data will be delivered must be generated at the source station prior to the protocol working.Most protocols tend to choose the 'delivery ratio' as an indicator instead of network goodput.As such, all the factors that have a significant impact on data delivery should be considered as advanced features.Rather, few protocols offer complete solutions for information redundancy.
The following issues centre around how the criteria are selected and what additional information is required in VANET.
More than two criteria are considered in several schemes for better simulation.For example, CAR introduces both contextual and link status information, and it can then redefine the concept of contexts.The network topology, link quality, and geographic location are incorporated into TLG to favour the optimal dissemination in VANET.The idea is to create a repository of contextual elements that integrate to form a cohesive whole.Of course, being focused on a single specific element will not be available, but selecting too many elements is a disservice.
Schemes for data dissemination usually aim for higher delivery ratios and shorter delivery delays, without considering their possible failures caused by transmission quality.We have found that the simulation results will be more credible when the network goodput is introduced.Although it is possible for vehicles to pursue data delivery with one type of realistic geographic information, focusing on road topology seems more important.In addition, as the vehicular nodes have their own unique movements, it might be helpful in complex traffic environments to launch a IET Intell.Transp.Syst., 2018, Vol. 12 Iss.10, pp.1189-1200 This is an open access article published by the IET under the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/)discrete, optimal dissemination with the trajectories of the nodes.Table 4 gives an overview of mobility models available for data dissemination in VANET.The more elements there are in a model, the greater the precision of emulation is and the more truthful model.With an overall analysis of all traffic sector conditions, it might be possible to determine the right model.The 'street networks' in Table 4 provide a vital factor in regulating vehicle movements, where the length of a road, the number of crossroads, and the density of buildings pervasively influence the traffic flows.The 'S-D' stations illustrate the spatiotemporal mode that the traffic flows exhibit.The various barriers predefined in the network are introduced into the strategies of 'traffic control' to emulate vehicle movement towards a stop sign or a traffic signal.Table 4 indicates less emphasis on types of vehicles and drivers' behaviours for the existing work, which makes it difficult to adapt to any surroundings for a mobility model.This factor might motivate future direction for introducing human factors in data dissemination, and in fact, most of the key players in that field have already envisioned this next step [57].
Selecting the right number of hops is essential.Due to uncertainty in the way vehicles move and the public traffic environment in which the vehicles operate, it is difficult to identify a set of prospective forwarding nodes.As such, the emphasis is on that which treats the next carriers, continually determines the appropriate hops and tests them to ensure that the TIS-related data delivery always remains effective.

Simulation study of data dissemination
With so many key points highlighted in this paper for designers to improve protocols, data dissemination remains a complicated topic from a cognitive standpoint.Although solution has improved, we can only evaluate the performance of these protocols relative to certain perspectives.Although it is crucial to test and evaluate protocol implementations in real testbed environments, it is difficult for designers to organise such large numbers of vehicles for network routing.Therefore, simulation is the mean of choice in the validation of disseminating protocols for VANETs.
We provide in this section a brief description of the simulation infrastructure federating two independent simulators, which provide features for vehicular motion modelling and networking capabilities.In addition, a critical aspect in the simulation study is the need for identifying simulation-related indicators reflecting the real behaviour of the traffic environment.

Vehicular networking traffic simulation
To be of use to the vehicular networking community, these protocols need to be made available with the vehicular traces and communication generated.In this section, we illustrate the need to select the most appropriate simulators and give a description of the application of these disseminating protocols to the vehicular networking scenario by creating an interaction between two simulators.

Network simulator choices:
Network simulators play a key role in these evaluations.It is necessary to determine a simulator that provides communication services to exchange data and synchronise the computation for dissemination.
A state-of-the-art network simulator, QualNet, applies only to emulate the wireless network.It contains a large set of network protocols, which improves its modelling capability, especially the increased simulation velocity and precision.Due to its management of multiprocessors, QualNet makes wireless simulation easier to perform for end-users.However, it is not available free of charge and cannot emulate wired communication so that QualNet cannot expand beyond specific uses.The Opnet is also a commercialbased network simulator and is available in various assemblies, including Modeler, TGuru, SPGuru, and WDMGuru.Modeler remains the most powerful for colleges and universities worldwide and has three levels of models, which correspond to the real internet infrastructure.Unlike QualNet, Opnet appears to have been designed with the strengths and uses of discrete events for network modelling.Moreover, the Opnet specialises in statistical studies on simulation results.However, supporting simulation in Windows only is one such potential downside, and the outcome of simulation will be influenced along with the network expands.The most widely used simulator is NS2, available as Open Source in the study of wireless simulation, resource allocation, and communication protocols.It is simple and efficient to emulate the major layers of the OSI stack and the major protocols for MANETs.The modularity and universality of NS2 is plainly seen at scalability and provides complete functional support as well as high-precision, high-speed, and efficient simulation.Although it is popular, NS2 was not originally designed to fit on graphical-user interfaces (GUIs).Thus, the user can only program the modules to construct an image.
In consideration of data dissemination in VANET, NS2 seems the most adapted simulator.Specific reasons could be that it is easier to extend by adding extra C programs for modelling of vehicular network; it has been used over decades worldwide in VANET simulation by an extensive number of researchers, which has resulted in extensive references for further evaluations; it allows the transmittal of pre-generated traces to NS2 directly, without much control over how the traces are created.

Decisions on traffic simulators:
Most of traffic simulators, such as VISSIM, claim to provide realistic vehicular mobility at both the microscopic and macroscopic levels.However, the generation of useable movement by networking simulators must be considered when decisions are made.
The VANET mobility simulator (VanetMobisim) is a scalable traffic simulator coded in Java, which, with its own platform and simulator, aims for greater consistency with realistic traffic environments.The VanetMobisim facilitates federating a network simulator, such as NS2 and GloMoSim, and enhances the level of realism.For example, it can macroscopically model a street network, one or two-way driving, speed and category limit, and traffic lights.At the same time, the built-in widely used models (intelligent driver, lane changing) also provide the benefit of providing details of each vehicle, e.g.perpendicular motion and longitudinal acceleration.The downside to the VanetMobisim is that it lacks specificity in general situations because of the platform-independent and extensible modules provided.
The free traffic simulator SUMO has been validated based on real traces.It aims to model contextual information, having the properties of space continuity and time discretisation.Similarly, SUMO can also indicate path-finding dynamically for a large scale of road networks supported from both macroscopic and microcosmic aspects.Due to the efficient GUI provided, SUMO provides quicker command methods.At the same time, it is possible to create many network configurations, including data formats and procedures.However, SUMO can only generate vehicular mobility traces in cooperation with MOVE and TraNSLite.The commercial traffic simulator, named as VISSIM, loads onto Windows and consists of vehicle actuated programming modules.One aspect that stands out is that VISSIM exemplifies traffic situations in three dimensions.Owing to bus-priority rule [58], changes of traffic flow can be more easily observed and analysed.In addition, lots of controlling indications are simulated.It is convenient for automatic car-following, lane-changing, and path-planning.The problem with VISSIM is that it places the needs of the models generated before those of simulations, subjecting such models to format the traces.It is also difficult to create effective interactions between VISSIM and other network simulators.
Although more complex interactions need to be learned to use VanetMobiSim effectively, VanetMobiSim is now a wellestablished simulator.It can fulfil minimum requirements for a realistic modelling of vehicular motion patterns [59], partly because it can connect with the network simulator, more specifically, the traces generated can be used by NS2.

Simulator-related criteria
The vehicular network scenario can be constructed with the enduser by visualisation means, which also is a key capability for the simulators.We, therefore, provide the following additional criteria applied for both traffic and network environments: • Simulation area -Predefined geographical area, which is always regarded as a rectangular area.Also, when considering the constraints on simulators, we do not only take into account how the scenario can be built; however, we provide guidelines for assessing through the use of various indicators.We, therefore, define the following criteria: • Transmission capacity -Indicates the transfer rate for the information, which is expected to reach the destinations in the application layer.all the predictions according to build-in models.• Density effect -Features such as delivery ratio and messageslost ratio; it has been aggregated with different densities to identify the appropriate range of density for each protocol.• Speed effect -Features such as end-to-end delay and information-lost ratios; it usually combines with different travelling speeds of vehicular nodes to analyse the proper range of speed for each protocol.• Network aggregated throughput -Messages accessed by all the nodes in the simulation process.• Network aggregated goodput -Messages correctly accessed by all the nodes in the process.

Discussion
It is necessary for the network simulator to load the mobility patterns, e.g.traces generated from traffic simulators.Therefore, the interaction between the two simulators is essential.As Harri et al. illustrated, federating already existing and validated network and traffic simulators will serve to provide advanced features for vehicular motion modelling and networking capabilities.From what has been shown in this survey, our opinion is that it would be advantageous to collaborate VanetMobiSim and NS2.We propose in this section to assess how to conduct the interactions as well as the new challenges this entails.
To allow a direct interaction between the two simulators, the mobility scenario extracted from the VanetMobiSim, including the road network topology, vehicle status, and its positions, has been transmitted to NS2.Both simulators work in parallel and thus might require interaction by transmitting the current positions and instructions.However, questions to address include what the sufficient requirements of communication interface between both simulators are.A significant obstacle to verifying is how mobility patterns can be altered by radio links, and vice versa.As such, this topic might guide a future direction for designing data dissemination.

Conclusion
VANETs characteristics include higher mobility and a limited degree of connection, which makes traditional data-dissemination protocols inefficient or unusable.A key aspect when transferring the requirements of TIS opportunistically is the use of intermediate vehicles that serve as store-and-forward message switches.In this survey, we introduced the challenges for service discovery, in particular for data dissemination.We then covered approaches to routing scheme and dissemination protocols and classified the most popular protocols available according to the indicators proposed.Better dissemination protocols can improve the efficiency of TIS, making it more convenient for travellers to obtain information.However, many challenges remain in designing protocols in modern VANET scenarios, such as improving network goodput, describing the mobile traces of vehicles, selecting the best forwarding number of times and nodes in one decision.We also described the simulation study to collaborate the traffic and network simulators to create an online reliability evaluation system.Insight into the future challenges in joint the VanetMobiSim and NS2 is finally provided.
This study aimed to improve understanding of the emerging developments of dissemination protocols for TIS discovery, including different methods related to various criteria and their evaluations.We anticipate additional input to influence and guide, with the understanding that this research is a starting point in determining future potential contexts of use.

Fig. 1
Fig. 1 Framework for the generation of data dissemination

Fig. 3 Fig. 4
Fig. 3 Finding the best candidate towards D

•
Simulation time -Length of time throughout the simulation process.• Transmission range -Longest radius for a vehicular node to transmit a message every time.• Vehicle speed -Average velocity extracted from the vehicular mobility patterns, which can be reset in the process depending on specific application requirements.• Packet rate -Transmission rate of every packet.• Node density -Average number of vehicular nodes in a specific road segment, which has situations similar to the above.• Mobility model -Selected from many mature models to achieve reliable simulating results.• Packet size -Size for storing a complete packet.• Number of nodes -Quantity of nodes involved in the process.• MAC protocol -Always set to 802.11 DCF.

•
Transmission delay -Time spent transmitting messages from a source to a destination.• Message-exchange capacity -Number of exchanging message between nodes under a particular network energy and bandwidth.• Buffer occupation -Requirements for storing space for each transfer point to ensure the storage of complete routing tables.• Message loss -Number of messages generated by the source stations that cannot reach their destinations.• Replicas -Number of replicas involved in the situations of different vehicle densities.• Broadcasts -Number of broadcasts to inform all the interested nodes during a period of time.• Hops -Number of intermediate nodes where a message is delivered from a source to a destination.• Level of prediction -Ratio of correct predictions compared to

Table 3
Constraint criteria in popular data dissemination