A comprehensive approach to avoid node selfishness and data redundancy

Abstract

Delay Tolerant Network (DTN) is an array of protocols which work together to enable a standardized way to communicate between nodes is store-carry-and-forward approach. In DTN, nodes help to convey message even with connectivity problem. Instead of using the own resource to pass a message to other nodes, a relay node is preferred which discovers the sink node with a greater probability of transferring messages on its own. A condition named node’s selfishness is thus raised and it surely degrades network performance, which is more obvious in urban environment. To handle such problem, an incentive-based routing algorithm is generated by using the concept Selfishness-Enhanced Reliable Forwarding (SERF) to remove node selfishness and Direct Diffusion to omit duplicate messages present in relay node. This algorithm finds the probability of messages received by the relay node with reference to the usage of resources of sender node. Concurrently, buffer management policy is maintained, by setting the threshold of message copies according to the resource consumption of the source node when it generates a message. The result reveals that the proposed algorithm is better to the existing algorithms with regard to the performance metrics such as, delivery ratio, the average delay, and network overhead.

Keywords

DTN buffer management incentive mechanism Direct Diffusion SERF

1 Introduction

Delay Tolerant network is a sparse as well as an intermittently connected mobile adhoc network; whereas message transmission is carried out in the absence of reliable communication and end-to-end connectivity [2]. In DTN, a type of network approach called “store-carry-forward” is introduced to transmit messages Voyiatzis and Artemios G [5]. Since, there exists no start to finish communication between node pairs, higher delay occurs and therefore DTN is used i.e. while transmitting message from the source node, if it can’t transmit to a sink node during that period, the node will not delete the message instantly, instead it will be stored in buffer of that particular node until discovers the sink node to receive the message else other relay nodes will transmit the message Tornell et al. [3]. For this, the message delivery ratio, delay, and network overhead must be considered while transmitting messages. The routing algorithms have the major advantages with reference to simplicity, its usage and customization.

Many routing algorithms [6 –11] have been proposed earlier to improve data delivery, reliability. These algorithms implicitly assume that all nodes are willing to forward packets for others. They have been designed without considering users’ willingness. But their always arises an issue named node selfishness ie) a node will not forward its packet to others in order to consume its resources. Then, previous algorithms may not work well since some packets are forwarded to nodes unwilling to relay, and will be dropped.

Routing in DTNs actually not a simple task as it works on the packet replication policy of store-carry-forward paradigm. Routing in DTNs follows packets replication method instead of packet forwarding in conventional network. Replication can be a serious problem, if replicated packets are more than a certain limit Wang et al. [11]. The node’s selfishness is more evident in the urban environment, instead of using the own resource to pass a message to other nodes, a relay node is preferred which discovers the sink node with a greater probability of transferring messages on its own since nodes have sociality. Chen et al. [4] This selfish behaviour of nodes are actually to save their resources, because if any relay node will participate in relaying process, they have to spend energy, buffer etc if they will not participate, they can save it. That is why some times they deny to carry packets. Thus, the node’s selfishness will notably degrade the network’s performance. To deal with this type of problems an incentive approach has to be proposed in DTNs.

To solve node selfishness, relaying process is used and thereby data are diffused throughout the network and duplicated initially. In DTN, nodes diffuse data when they are in contact. Because the contact duration between two nodes is pretty short and the buffer size of each node is within certain limit. Thus, the transferred data has to compete for the limited bandwidth and buffering space. Therefore, the diffusion decision made by each node such as which data should be propagated first and which data should be replaced first, affects the diffusion speed and the data access delay.

Message replication can allow for substantially better message delivery ratios than in forwarding-based protocols and the protocols allow an original message to be replicated. Though it is better, their arises some issues such as network congestion in clustered areas and being wasteful with network resources including bandwidth, storage, and energy. Since network resources may quickly become constrained, deciding which messages to transmit first and which messages to drop first play critical roles in many routing protocols Wang et al. [6]. Therefore, to solve this issue Data Diffusion (DD), a flat routing-based protocol is used which omits duplicate messages. The major idea here is, DD makes the decision about which replicate message to transmit at first and which one to eliminate.

Nodes must be ready to participate in relaying process and thus selfishness can be eradicated which leads to improve the routing parameters like delivery probability, average delay, network overhead. In the existing system, wastage of energy occurs as the duplicate message spread all over the network and one must wait till it reach the subscriber. Also, reward is provided only to the selfish node to make it participate in the message forwarding path Wang et al. [11]. Some nodes adamantly behave selfish manner even after providing reward too. In our research work we have used incentive-based routing strategy for node selfishness along with duplicate message elimination. A SERF, Incentive based routing strategy for DTN is proposed. An incentive system, that considers variables such as message frequency, interest level, battery use, etc. for calculating incentives. Incentive mechanism is the process to improve the performance of the routing protocol. The source offers a reward for each and every relay it encounters, but tells them that there will be a reward only for the first one to deliver the message to destination. The incentive a relay asks for has to cover its expected cost, as the relay determines when it reaches the source. In SERF selfish nodes, those nodes which do not want to transmit packets to destination is removed from the forwarding strategy and incentive is provided to the relay node by making it act as the source node i.e. relay node participate eagerly without considering resource consumption. By using this mechanism relay can earn fair incentives based on their packet replication Wang et al. [8]. For example, if the source tells a relay that many copies of messages are already in circulation, the relay would obviously predict a lower probability of receiving the reward (because other relays could have already delivered the message), and thus this relay will naturally ask for a higher reward to cover its expense. So based on the reward value the incentive is calculated. The proposed work is implemented in ONE simulator and found that delivery probability has increased and moreover delay and overhead has reduced, this shows that our proposed mechanism is working properly for socially aware DTNs.

The other sections are as follows. In section 2 related works are given, the model of the network and the proposed work is explained in section 3, while the results and discussions are described in section 4. Finally, in section 5, the conclusions are outlined.

2 Related works

Traditional network resource and end-to-end data delivery performance networks are not any more effective in continuously increasing traffic. Further some relays don’t transfer the message to other nodes since message delivery exhaust resource such as storage and power. Thereby, a mechanism is developed to stimulate the node and transmit the message to overcome the misbehaviour of selfish nodes. A DTN is a portable ad hoc wireless network, where end-to-end paths are not existing at any specified time or is difficult to maintain. Message delivery is premised on opportunistic mobile node encounters and is rendered in the fashion of “store-carry-forward”. Conversely, it has its drawbacks, such as afflict the buffer and the inefficient usage of contact duration. The routing protocol is a vital area which deduce the DTN performance. Although DTN is disclosed to be tolerant of delay, protocol shows better result that will maximize packet delivery rate and minimize the delivery latency [12, 13]. Various routing algorithms have already been suggested to handle such attributes of the network they are, epidemic routing and probabilistic routing. These routing algorithms are based on the concept named multi-copy routing. During the process of transmitting or storing messages, these type of routing takes large quantity of resources from both network and buffer, leading in a rapid depletion of network resources and succeeding with poor performance of network [14]. In urban environment, this type of poor performance in network is more common which leads to lacking consideration for another node while delay tolerant is applied. Consequently, designing a routing algorithm to solve node selfishness is a significant goal to achieve.

Selfishness. Vahdat et al. [15] proposed the Epidemic Routing Protocol, which is similar to the flooding method of spreading messages. Epidemic routing transmits messages based on subsequent manner, the message carrier moves in the network, and when it meets another node with no replica of the message, it will be transferred to the node during the contact duration. In other terms, when two nodes meet, they compare the summary vector of each other’s message, then interchange any messages that the other does not have. If the space of the buffer in a node is sufficiently large, epidemic routing provides greater delivery ratio and little delay in delivery. Though, space for storing message in a node is restricted in real network, this algorithm will generate a lot of duplicate messages in the network. This will result in large overhead in both communication and network congestion.

Chen Yu et al. [17] In the PRoPHET protocol, every node has a routing table that includes the probability of this node routing to any other node in the network. It is a helpful technique for solving the sparse Vehicular Ad-hoc Network (VANET) issue. Ota et al. [18] described about a correlation-based approach in urban city area to convey the message between vehicles. Compared with epidemic routing, this protocol does not distribute the data to all the encountered nodes, instead it makes decision about whether to transfer a chunk of data to the identified node by comparing the probability value to the sink node. So, this technique could significantly eliminate the probability of message transfer interruption, and thus it could upgrade the message’s delivery probability and notably limit the network overheads.

Guizhu Wang et al. [16] proposed an improved routing algorithm which is on basis of node’s congestion level (PROPHET-CLN). The traditional Prophet routing algorithm only takes the probability of message delivery into account without taking care of each connection time. Besides solving the problem of node selecting based on delivery probability, it also assigns the right size of messages in relation to the node’s capacity. This algorithm has better message delivery rate, network overhead and average latency but improvement must be made in effectively control congestion.

Direct Diffusion Routing is a data centric approach. Direct diffusion method has greater energy efficient technique by reason of having no requirement to maintain the global network topology. Raja. M.S and Poornima. T [23] described a comparison of flat routing protocols. Here, survey is done on Flat Routing Protocols namely, Sensor Protocols for Information via Negotiation (SPIN) and DD. DD prevail against the issue of SPIN of reliable delivery. SPIN protocol is not scalable because if the sink is not eager to generate too many events, their energy could be minimized by the sensing nodes around it. However, DD is more scalable than SPIN. Directed diffusion has better throughput compared to SPIN though it has excellent reliable delivery.

Khalid S and Zeyad T [24] discussed on data centric based protocols for routing. Directed diffusion allows the path change and repair. If a path hasn’t gone well, another path is assigned. Therefore, there should be a plenitude of paths from the launch of the task. Directed diffusion includes communication that is neighbor-to-neighbor. Since each node can sense, cache and aggregate, node addressing system is not needed. Here caching helps both the energy efficiency and the delay. Direct Diffusion is energy efficient because of its on-demand feature. The problem with directed diffusion is that we cannot apply it to all the applications of the sensor network. This is because it has been based on a data delivery model that is query-driven.

Sobin [19] presented a novel buffer managing scheme based on hop-count and TTL. To obtain better efficiency, the restricted encounter probability must also be used efficiently by routing schemes in DTN’s. Prioritization of message must be done in node’s buffer so to transfer the message scheduling must be done and optimized performance is transferred first. Logical partitioning of the buffer is done based on hop-count value and threshold. Message with lower hop-count value than the threshold is sorted according to their hop-count and message having higher hop-count value than the threshold is sorted based on their Time to Live (TTL) values.

An incentive scheme based on multi receiver named MuRIS is presented by Gholap et al. [20]. Here, the scheme MuRIS provides each node with a recompense i.e. as a stimulation for each node to transmit the messages to subsequent node. An incentive mechanism is presented by Brun et al. [21] for DTN. Here, only the relay that is the first to shift the message to the sink recompensating is done. Seregina et al. [22] suggested a recompense mechanism during DTN transmitting operation where relays lose their remembrance and accumulator.

Qaisar Ayub and Sulma Rashid [23] proposed a delay tolerant network routing protocol called resource refrain quota-based routing protocol. The proposed protocol assigns the energy transmission quota to the number of copies of messages ’ n.’ In addition, message was transmitted only to those nodes which had a high probability value to reach the destination of the message. Yuxin Mao et al. [24] proposed a new routing protocol, called the History of Encounters and Transitivity (PROPHET) scheduling-probabilistic routing protocol. In this protocol, we measure the predictability of the distribution according to the frequency of encounter between nodes. Two scheduling mechanisms are proposed to expand the standard protocol of PROPHET and boost both storage and transmission capacity in DTN.

Xuebin Ma et al. [25] proposed a reliable energy-aware routing protocol, called RER. A hop-by-hop retransmission recognition mechanism is introduced in the RER to ensure the reliability of message transmission. Second, to support RER, develop a metric called Reliable Distance-Based Energy Cost (RECBD). Sweta Jain and Ankit Verma [26] presents a new incentive scheme to discourage selfish behaviour of nodes in delay-tolerant networks. To discourage selfish behaviour of nodes and to motivate them to forward the messages of other nodes, a credit-based scheme has been applied. A hybrid technique is proposed which combines both the incentive- and reputation-based scheme. The nodes behaviour is monitored by observing the number of messages forwarded and dropped by them to detect when a node starts acting selfishly. A credit-based incentive scheme is then applied to motivate nodes to avoid their selfish behaviour and cooperate in message forwarding. The credit rewarded is a function of the message size and the remaining TTL of the message.

When many messages are forwarded to the nodes with high connection degree, these nodes will become congested and deliberately discard messages, which will seriously degrade the routing performance and reduce the benefits of other nodes. To address this problem, Qingfeng et al. [27] proposed a credit-based congestion-aware incentive scheme (CBCAIS) for DTNs. In CBCAIS, a check and punishment mechanism are proposed to prevent forwarding nodes from deliberately discarding message. In addition, a message acceptance selection mechanism is proposed to allow the nodes to decide whether to accept other messages, according to self congestion degree.

Meanwhile, to figure out the trouble about nodes’ selfishness present in the network, Mao and Zhu [28] proposed a routing algorithm with regard to energy. In which more messages are transmitted by the node to obtain more services from other nodes. By considering the energy resources, the algorithm can efficiently enhance the network’s efficiency. Here, the node’s buffer resources are not taken in to consideration. Moreover, the algorithm doesn’t take care about how to control the message replicas in the network.

From above analysis, an incentive mechanism is developed for DTN based routing. The contribution of this work is as follows:

(1) An incentive mechanism is proposed to eliminate node selfishness. By this method, a reward is provided to the relay node by making it act as a source node i.e. it participates in the forwarding path instead of a selfish node.

(2) A buffer management technique is intended to provide a solution to the problem of using extremely limited node resources efficiently premised on the value of contribution function.

(3) Direct diffusion protocol is proposed to omit duplicate messages.

3 Network model and proposed method

3.1 Network model

Network model is one of the best way of representing the nodes and their relationships. The network structure will be classified in line with node uniformity. The nodes in some networks are deployed uniformly and be adequate to one another. Specifically, they route the knowledge supported the networks design. This address two styles of node deployments, nodes with constant level of association and nodes with totally different hierarchies. In this delay tolerant network, the problem of selfishness occurs between the two edges of node at time t. One of the process to solve this problem is to find the contribution value of the node. The computation of the contribution value is explained as follows:

Take node N within the network to take care of a listing, $V (N) = {A_{t} (N), R_{t} (N), Δ E_{t} (N), Δ B_{t} (N), RC (N), P_{N}^{M}}$ . A_t (N) is that the contribution value of node N at time t, as given in Equation (1); Node N consumed the resource at time t is denoted as R_t (N) and that value is given in Equation(2); Node N consumed the energy at time t is denoted as ΔE_t (N) and also that value is given in Equation (3); Node N consumed the buffer value at time t is denoted as ΔB_t (N) and also that value is given in Equation (4); Node N forwards the messages to relay nodes or destination nodes which is record by the matrix represented as RC (N) is given in Equation (5); The probability of node M encounters with node N is represents as $P_{N}^{M}$ . V(N) represents the list of values taken in the network by the nodes. The probability of alternative node is recorded by the every node N which is given in the following section. $A_{t} (N) = (1 - α) \times A_{t - 1} (N) + α \times R_{t} (N)$ (1)

Here, α= adjustment factor

A_t(N) = Contribution value of node N

R_t(N) = Resource Consumption of node N

R_t(N) is given as: $R_{t} (N) = β \times Δ E_{t} (N) + (1 - β) \times Δ B_{t} (N)$ (2)

Here, β = adjustment factor

ΔE_t(N) = Energy Consumption of node N

Δ B_t(N) = Buffer Value of node N

Δ E_t(N) and Δ B_t(N) values are derived by using Equation (3) and Equation (4), respectively. $Δ E_{t} (N) = \sum_{0}^{t} ɛ^{N} (E_{c}, t)$ (3) $Δ B_{t} (N) = \sum_{0}^{t} ɛ^{N} (M_{c}, t)$ (4)

Where, E_c = Energy Consumption M_c = Buffer Consumption

In equation (1) the contribution value of the node n at time t is derived by multiplying the values of adjustment factor, the contribution value at time t-1 and also the resource consumption value. For calculating the contribution value for the node, need to find out the resource consumption value. To get the resource consumption value, multiplying the energy, buffer values at time t and adjustment factor. Which is defined in equation (2). The equation (3) and equation (4) explained the buffer values and the energy values which are calculated by adding all the energy and buffer consumption values at all the time intervals.

Here the event statistics function is denoted as ɛ. The energy consumed event is expressed as E_c. The buffer usage event is M_c. $RC (N) = [{RC}_{N}^{1}, {RC}_{N}^{2}, . . . . {RC}_{N}^{m}]$ (5)

Number of times the source node sends the message m to destination node is denoted as ${RC}_{N}^{m}$ . List of attributes maintained by each message m within the network is denoted as L(m). L (m) = (S_m, D_m, C (S_m, N_m)), wherever the source node of message m is S_m. The sink node of message m is D_m. Once the message is generated, the contribution value of the supply node is C (S_m). Number of copies of the message m is N_m. C(S_m, N_m) is the contribution value for the message copies and source node. When the number of copies of the message is bigger than one, then the message m solely occupies one message buffer area of node.

3.2 Proposed method

An SERF routing algorithm with DD protocol mechanism is proposed to solve the node selfishness, eliminating redundant data and also shrinking range of information transmission problems in the DTN. The block diagram for the proposed routing algorithm is illustrates in Fig. 1. In this proposed work the message is forwarded to the source node. The source node contains the contribution value and the copies of the message. After this, the message is transferred to the buffer management system. Where the messages are stored using buffer storage mechanisms and the priority of the message also controlled here. In this buffer management system, the copies of the message are controlled and the node selfishness also adjusted. By using the forwarding strategies of the buffer management system the message is forwarded to the destination node. After using all the above mechanisms of the proposed work the message is transferred to the destination node without any delay.

Fig. 1

Block Diagram for the Proposed Algorithm.

The working method of SERF routing algorithm’s routing method and the buffer management policy with selfish node management in DTN are explained as follows:

3.2.1 SERF algorithm

In this algorithm, the message is transfer to the sink node. The contribution value is used to determine the likelihood of messages received while transmitting messages from relay nodes. The number of copies of the message will be set when it is generated, based on the contribution value. This calculated contribution value is used to find the probability of the received message. The following routing method explains the probability calculation of the message.

3.2.1.1. Routing method.

Step 1: It is assumed that node N has messages that need to forward to Node M.

The remaining energies of node N and Node M are Δ E_N and Δ E_M, and also the remaining buffer size of both nodes are Δ M_N and Δ M_M.

Step 2: Within the buffer of node N, calculate $η_{N}^{m}$ and $η_{M}^{m}$ for any message m using Equation(6) and here Node N establish a connection to Node M. If $η_{N}^{m}$ < $η_{M}^{m}$ , then the node N forwards the message m to node M. P (A_t (N)) is the probability of Node N sends the message m to Node M and this probability is calculated using the Equation (7). $η_{N}^{m} = \frac{{RC}_{N}^{m}}{{RC}_{N}^{total}}$ (6) $P (A_{t} (N)) = 1 - \frac{1}{e^{γ c_{t} (N)}}$ (7)

Node N sends the message to the destination node of m based on time is ${RC}_{N}^{m}$ .

The total value of ${RC}_{N}^{m} is$ ${RC}_{N}^{total}$ .

Here, r = adjustment factor for routing based on time t and $e_{t}^{rc (N)}$ = exponential value for the routing time of Node N.

In this calculation, adjustment factor r is used. Adjustment factors are a number attribute that is applied or used in conjunction with nodes and message in the networks. The greater probability message only forward to the destination node. Before transferring the messages to the destination node, all the messages are should stored in the buffer space. This buffer management system contains the replicated message and the selfishness problem. This may cause delay in message transferring. To solve this problems in delay tolerant network, the selfishness node management scheme SERF and also the DD protocol is proposed.

3.2.1.2. Buffer management in delay tolerant network.

Due to very limited buffer space available in DTN nodes, buffer management is a very important factor in DTN. Although a scheduling policy in DTN determines which message to forward first, the drop policy decides which messages to drop in the event of buffer overflow. There are variety of buffer management schemes that may be adopted by varied DTN applications. To solve the difficulty of very restricted network resources, buffer management mechanism is planned. It can be used with efficiency. When the message is generated, the quantities of copies allotted to message m is $C_{m}^{t}$ for the supply node of message m.

The $C_{m}^{t}$ calculation is described in the following Equation (8): $C_{m}^{t} = K \times A_{t} C (S_{m})$ (8)

The calculated value for the message copy quantity is derived by using the delivery chance value (K) and also the contribution value for the source node at time t (A_tC(S_m)).

Suppose that M be the set of all messages created within the network and Nd be the set of all messages delivered. Then, $deliverychance (k) = Nd / M$ (9)

Overhead quantitative relation is outlined as what percentage reproduction packets are forwarded to deliver one packet. Then, overhead quantitative relation is outlined as: $(Number of total forwarded message - Nd) / Nd$ (10)

Let the i^th delivered message was created at time Ti and delivered at time di. Then, average message delivery latency= $(\sum (di - Ti) i = 1) / Nd$ (11)

In buffer management algorithms, the way to management the buffer area occupation is incredibly key. Here the tendency is: $\frac{S_{i}}{W_{i}} = \frac{S_{j}}{W_{j}}$ (12) where S_i is that the buffer area occupation and S_j is also the buffer space which was already occupied by the source node message. Once the cache is full, the service flow with the most important price of Si /Wi are going to be dropped so as to ensure fairness. Here the fairness is reflected in packets with completely different queue length. Assume that m_i is that the weight, and n_i is that the current queue length of the service flow i. The artificial weight State will be calculated as: $W_{i} = α \times m_{i} + (1 - α) \times n_{i}$ (13)

Where α is that the modify parameter of the two coefficient. Coefficients mi and ni. α will be pre-assigned, or determined in accordance with usage of the cache. mi is expounded to the service flow itself, and completely different service flows are allotted with completely different weight values. As long because the service flow is active, this issue can stay unchanged.ni is time variable, which reflects dropping scenario of this service flow.

When the buffer overflows, the copies of message within the buffer is discarded. If the message copies which is generated by the source node are discarded then it has smallest amount of contribution value.

3.2.1.3. Selfishness node management in DTN.

Within the message forwarding some risks occur. The major risks occurred in DTN are node selfishness and data redundancy. To prevent from this risks, the incentive mechanism is applied. Using this concept, all nodes will store in their memory a list of topics/interests wherever they want. If the opposite node have an interest to one or additional of those topics, then they will unfold the subject. This behavior is called selfish behavior and it will not share the fascinating topics within the network.

Degree of position for a node i throughout an interval t is computed as follows: $E_{Degree} (i, T) = \sum_{j = 1}^{M_{i}} e (i, j, t)$ (14)

Representing a position between node i and node j on the graph similar to the slot. The node’s degree value over a collection of T slots is denoted as C_Degree, and it is described in Equation (15): $C_{Degree} (i) = \frac{1}{T} \sum_{t = 0}^{T} E_{Degree} (i, T - t)$ (15)

The forwarding activity of every node is defined in the forwarding list. This list contains the list of nodes, message ID for each source and destination nodes and the time at once the message is received. Check whether the nodes within the network have selfishness or not. If they detect any selfishness, then they increase the selfishness score. Otherwise, decrease the selfish score. In this incentive mechanism, the threshold value is fixed. The message will not be received when the selfishness score is larger than the threshold value. From this, the selfish node is detected. Because of this behavior, the node consumed more energy. So this selfish node is discarded from the forwarding path. At the same there will be some problem occur in message forwarding. For that, the relay node is acting as the source node to forward the message. After solving the selfishness problem, have to omit the replicated messages in the buffer space. For that purpose, introduced one flat routing protocol named as Direct Diffusion routing protocol.

3.3 Flat routing protocol

The flat routing protocol algorithm is to achieve the aim of eliminating redundant data and shrinking range of information transmission. In this protocol all the nodes within the network play constant role. Flat specification presents many benefits, together with lowest overhead to take care of the infrastructure between nodes. The proposed Flat routing protocol is called as the DD protocol. It has three phases. They are explained as follows: In path establishment part establish best path (minimum hops) between supply and destination. This part have minimum overhead and minimum information measure consumption to deliver the messages in a very timely manner. In data transmission phase the information is transfer from source node to the destination node. In Path increase phase, send increased messages to the neighbor nodes exploitation the quickest speed, and eventually selected the most effective path between the supply node and also the target node.

3.3.1 Direct diffusion (DD)

In DTN, DD protocol is designed for robustness, scaling and energy efficient. This protocol also provide less traffic than flooding and also transmits the data along the best path. In this proposed work it reduces the redundant data. This qualities lead to overcome the delay in DTN. It is one in every of the largely used and common information central strategies. Information central is wherever that work as associate application aware technique, it permits to pick out by trial and error best path that saves the energy by additionally caching and process the info in-network that’s it combines the data from completely different sources by eliminating the redundancy and minimizes the quantity of transmission. Because it is that the information central technique for the addressing mechanism as all communication is neighbor-to-neighbor.

The different ways ranging from a similar source are portrayed by a group S_path. $S_{path} = {[N 1, N 2, \dots Nn}$ (16) Each element N_i (1 ⩽ i ⩽ n) in S_path can be represented as follows: $< H_{11}, R_{11} >, < H_{12}, R_{12} >, \dots < H_{1 k}, R_{1 k} >$ (17)

Based on the S_path value it find the best routing path. This routing path contains the source node, neighbor node and also the destination node.

where H_ij (1 ⩽ j ⩽ k) denotes the node that has j hops from supply node S within the path Pi, R_ij (1 ⩽ j ⩽ k) denotes the residual energy of node H_ij and H_ik denotes the neighbor of the sink node. The neighbor node H_ik is also called as relay node. The data in the source node is compared with the data in the destination node. If the messages in the relay node is equal to the source node, then the redundant data will be discarded. The relay node is used to transfer the message to the destination node, when node selfishness and redundancy problem occurred. Because of this process the energy consumption is reduced. Since an iterative process computes successive approximations to a linear networking system solution, a realistic test is needed to determine when to end the iteration. Ideally this test would measure the distance to the true solution from the last iterate, but this is not possible. Instead, different other metrics are used which typically involve the residual. In DTN, the transport layers carrying the bundles across their local networks are called bundle convergence layers. Statistics is about collecting and analyzing data and making statistical projections and forecasts. In this paper, the convergence of contribution value and the average probability have been applied.

The algorithm for the proposed work is explained as follows: The aim of this algorithm is to solve the node selfishness and data redundancy problem. To solve those issues, it first take the inputs which are needed for the process. The inputs of this method are network size, node degree and also the buffer size. Here, the node degree defines number of nodes in the DTN and the buffer size described about the space allocated for the message. In this, first transmit the message to the neighbor node which should match on the contribution value. The neighbor node is denoted as node N. Based on the incentive mechanisms, should calculate the contribution value. If the resource consumption value is less than zero, then increase the contribution value and set the contribution value for the sender node. If the message copy is less than one, then it occupies only one buffer space. Here the message may be discarded because of low contribution value. Set the average probability value. Then find the probability of the message by using all the resource consumption, number of nodes and the number of forwarded message values. The probability value is used for the higher delivery ratio of message. Check whether the calculated probability is greater than the average probability or not. The higher probability message only forwarded. Also check the redundant messages in the relay node. After discarding all the redundant messages, it will forward to the destination node using best path without redundancy.

Routing algorithm for Node selfishness and Data Redundancy.

Algorithm Input: Network Size, Node Degree, Time, Buffer size

Output: Best routing path, Data without Redundancy

Begin

1. Transmit message to neighbor node

2. Match on contribution value

3. If (resource consumption<0)

4. Increase contribution value

5. Set contribution value to sender of Interest

6. if(message copies < 1)

7. message m occupies only one space

8. if(m = low contribution value)

9. discard the message

10. Find the probability P(M)

11. Prob P(M) = Resource Consumption+Number of Nodes+ ((Number of total forwarded message –Nd)/Nd)

12. Find average Probability P(M)

13. if(Prob(PM)>average ProbP(M))

14. Greater probability for forwarding message

15. Compare the received messages for redundancy

16. Message in relay node M_r

16. if (m = M_r)

17. Discard the redundant messages

18. else

19. Message forwarding to best path

20. Recieve the message

End

4 Results and discussions

The ONE simulator software is used for simulation. The results of proposed method are compared with the Spray and Wait, IRIS and MuRIS methods. Fig. (2) shows the one simulator environment with the routing process and the Fig. (3) illustrates the best path from source to the destination node. The node’s energy consumption can be described by equation (18).

$C_{b} = C^{Lis} + Cs + Cr$ (18) where C^Lis is the energy consumed by the listening channel. Per second the total amount of energy consumed is denoted as Cs. Total resource utilized for a particular channel is denoted as Cr. The energy consumption value for the particular node is derived by adding all the energy consumption and resource consumption for a particular channel. With the use of 25 nodes the node speed is increased at the rate of 370 kbps. Which is given at Table 1. The better communication will be done when the communication range is 6.32. Because within this range only, the message transmit without any interruption. When the message generation interval is 4.32, the message is generated without any redundant copy. So these values are set to get a better message forwarding in the network.

Figure 2 explains the routing process of the proposed method done in the delay tolerant network. The best routing path for message forwarding without any delay is described in Fig. 3.

Fig. 2

Routing process in one simulator environment.

Fig. 3

The Best path from source to destination.

4.1 Performance analysis

In this proposed work the performance is analyzed by using the three aspects delivery ratio, average delay, and network overhead. All these aspects of proposed method is compared with the other existing Spray and Wait, IRS and MuRIS methods.

4.1.1 Delivery ratio

The successful data received ratio is called delivery ratio. If the delivery ratio is high then is a better algorithm. At the sink node the number of messages delivered is termed as N_d(M). The number of messages transmitted by every source is termed as N_t(M). The delivery ratio of the message is described as follows: ${DR}_{m} = (N_{d} (M) / N_{t} (M)) * 100$ (19)

On the analysis of delivery ratio with the other existing methods the proposed method have the higher delivery ratio value. Using buffer management strategy, the messages are discarded properly. So the message delivery ratio is increased. The range of buffer size is between 4 to 20 kb. The delivery ratio of the proposed work varies between 0.61 and 0.78 based on buffer size. The range of time value is between 60 to 300 seconds. Based on time it varies between 0.74 and 0.79, that is considerably bigger than the three compared algorithms. The delivery ratio for various buffer size and time intervals are shown in Tables 2 and 3.

Table 1

Simulation Settings

Simulation time	114 sec
Node Speed	370 kbps
Number of Nodes	25
Buffer Size	8
Message Generation Interval	4.32
Communication Range of the Nodes	6.32

Table 2

Delivery ratio based on buffer size with existing methods

Buffer size	Proposed Method	Spray and Wait	IRS	MuRIS
4	0.61	0.52	0.32	0.48
8	0.67	0.63	0.34	0.55
12	0.68	0.62	0.38	0.54
16	0.75	0.71	0.39	0.65
20	0.78	0.75	0.42	0.71

Table 3

Delivery ratio based on Time with existing methods

Time	Proposed Method	Spray and Wait	IRS	MuRIS
60	0.74	0.52	0.32	0.48
120	0.76	0.63	0.34	0.55
180	0.79	0.62	0.38	0.54
240	0.75	0.71	0.39	0.65
300	0.74	0.75	0.42	0.71

The comparison graph for delivery ratio based on various buffer size and Time values with existing methods are shown in Figs. 4 and 5. Here the buffer size is varied between 4 to 20 kb and here proved that the proposed work have the high delivery ratio within a smaller amount of time. The time value for delivery ratio is varied between 60 to 300 seconds. Compared to other three existing methods, it provide better delivery ratio.

Fig. 4

Delivery ratio based on buffer size with existing methods.

Fig. 5

Delivery ratio based on Time with existing methods.

4.1.2 Average delay

The delay of a network specifies however long it takes for a small amount of information to travel across the network from one node or termination to a different. The average delay any given message is probably going to expertise is given by the formula $D_{m} = 1 / (R_{m} - A_{r})$ (20)

wherever R_m is that the range of message per second the ability will sustain and A_r is the average rate at that messages are incoming to be maintained. The average delay for each buffer size is listed in the following table and the value is between 4 to 20 kb.

The proposed methodology delay is tested with all alternative existing strategies. The proposed technique shows the best performance in terms of delay. Here the time required for the travelling of the data to destination node is highly reduced. The delay of the proposed algorithm varies between 2000 and 2850 based on buffer size and also it varies between 3700 and 3850 based on time, that is considerably lower than the three existing algorithms. Tables 4 and 5 shows the delay value for varied buffer size and time values.

Table 4

Delay based on buffer size with existing methods

Buffer size	Proposed Method	Spray and Wait	IRS	MuRIS
4	2850	3500	4300	3400
8	2800	2500	4100	2700
12	2700	2400	3500	2600
16	2500	2360	3200	2400
20	2000	2200	2800	2140

Table 5

Delay based on Time with existing methods

Time	Proposed Method	Spray and Wait	IRS	MuRIS
60	2850	3500	4300	3400
120	2800	2500	4100	2700
180	2700	2400	3500	2600
240	2500	2360	3200	2400
300	2000	2200	2800	2140

The comparison graph for delay primarily based on varied buffer size and Time values with existing strategies are shown in Figs. 6 and 7.

Fig. 6

Delay based on buffer size with existing methods.

Fig. 7

Delay based on Time with existing methods.

4.1.3 Overhead

The network load has been reflected as the overhead ratio. From the analysis of overhead values, the better algorithm should have a lower overhead value. The buffer size value is varies between 4 to 20 kb and also the time value is varies between 60 to 300 seconds.

The network overhead of the proposed method is tested with all other existing methods such as Spray and Wait, IRS and MuRIS. Based on the existing methods the proposed work have a low network overhead value. So, the network overhead is reduced. The overhead of the proposed algorithm varies between 15 and 30 based on buffer size. Based on time value it varies between 20 and 31, that is considerably lower than the three existing algorithms. The following tables (Tables 6 and 7) shows the network overhead for various buffer size and time values.

Table 6
Overhead based on buffer size with existing methods

Buffer size Proposed Method Spray and Wait IRS MuRIS

4 15 43 32 23

8 21 45 34 27

12 25 48 36 32

16 27 52 42 36

20 30 58 45 39

Buffer size	Proposed Method	Spray and Wait	IRS	MuRIS
4	15	43	32	23
8	21	45	34	27
12	25	48	36	32
16	27	52	42	36
20	30	58	45	39

Table 7

Overhead based on Time with existing methods

Time	Proposed Method	Spray and Wait	Irs	MuRIS
60	20	43	32	23
120	22	45	34	27
180	24	48	36	32
240	25	52	42	36
300	31	58	45	39

Figs. 8 and 9 shows the comparison graph for network overhead based on various buffer size and Time values with existing methods.

Fig. 8

Overhead based on buffer size with existing methods.

Fig. 9

Overhead based on Time with existing methods.

When compared with the other existing systems, the proposed work overcome the node selfishness and data redundancy problem with low delay and high delivery ratio. If the network have low delay value and high delivery ratio, then only it transfer the message within a small amount of time. So our proposed work have the greater possibility to transfer the message and also solving the issue of node selfishness and data redundancy problem.

5 Conclusion

The problem of node selfishness, eliminating redundant data and also shrinking range of information transmission are solved by using this proposed routing algorithm. This routing algorithm is easy and fair. These protocols have great amount of control packet overhead and low scalability. The proposed work have lower network overhead. So the copies of the message is easily reduced. On compared with the other three methodologies the proposed methodology provides the better network performance in terms of average delay, network overhead and delivery ratio. In the future, other aspects of the model, including security, will be done.

References

Wang

, Wu

, Wang

, Liu

and Zhou

, TCP convergence layer-based operation of DTN for long-delay cislunar communications, IEEE Systems Journal 4(3) (2010), 385–395.

Mangrulkar

R.S.

and Atique

, Routing protocol for delay tolerant network:Asurvey and comparison, In International Conference on Communication Control and Computing Technologies (2010), 210–215.

, Wang

, Yang

, Wang

and Wang

, Dynamic coding control in social intermittent connectivity wireless networks, IEEE Transactions on Vehicular Technology 65(9) (2015), 7634–7646.

Tornell

, Sergio

, Calafate

, Cano

J.-C.

and Manzoni

, DTN protocols for vehicular networks: An application oriented overview, IEEE Communications Surveys & Tutorials 17(2) (2014), 868–887.

Chen

, Shen

and Yan

, Multicent: A multifunctional incentive scheme adaptive to diverse performance objectives for DTN routing, IEEE Transactions on Parallel and Distributed Systems 26(6) (2014), 1643–1653.

Voyiatzis,

Artemios G

, A survey of delay-and disruption-tolerant networking applications, Journal of Internet Engineering, 5(1) (2012), 331–344.

Wang

, Lv

, Wang

and Feng

, DCAR: DTN congestion avoidance routing algorithm based on tokens in an urban environment, Journal of Sensors (2017).

Samaras

S.N.

and Foteini Triantari

, On Direct Diffusion Routing for Wireless Sensor Networks, In 2016 Advances inWireless and Optical Communications (RTUWO) (2016), 89–94.

Wang

E.K.

, Ye

and Xu

, Lightweight secure directed diffusion for wireless sensor networks, International Journal of Distributed Sensor Networks 10(4) (2014), 415143.

10.

Cai

, Fan

and Wen

, An incentive-compatible routing protocol for two-hop delay-tolerant networks, IEEE Transactions on Vehicular Technology 65(1) (2015), 266–277.

11.

Abdelkader

, Naik

, Nayak

, Goel

and Srivastava

, A performance comparison of delay-tolerant network routing protocols, IEEE Network 30(2) (2016), 46–53.

12.

Wang

, Wang

, Feng

and Lv

, NWBBMP: a novel weight-based buffer management policy for DTN routing protocols, Peer-to-Peer Networking and Applications 11(5) (2018), 917–923.

13.

Choksatid

and Prabhavat

, An epidemic routing with low message exchange overhead for delay tolerant networks, In Progress in Systems Engineering (2015), 429–436.

14.

Wang

, Tao

, Zhang

and Pan

, A improved Prophet routing based on congestion level of nodes in DTN, In 2017 IEEE 2nd Advanced Information Technology, Electronic and Automation Control Conference (IAEAC), (2017), 1666–1669.

15.

, Tu

, Yao

, Lu

and Jin

, Probabilistic routing algorithm based on contact duration and message redundancy in delay tolerant network, International Journal of Communication Systems 29(16) (2016), 2416–2426.

16.

Ota

, Dong

, Chang

and Zhu

, MMCD: Cooperative downloading for highway VANETs, IEEE Transactions on Emerging Topics in Computing 3(1) (2014), 34–43.

17.

Sobin

C.C.

, An efficient buffer management policy for DTN, Procedia Computer Science 93 (2016), 309–314.

18.

Gholap

and Shaikh

, Information sharing in delay tolerant mobile networks with some incentive and fewest transmissions, In 2016 IEEE International Conference on Advances in Computer Applications (ICACA), (2016), 197–201.

19.

Brun

, El-Azouzi

, Prabhu

and Seregina

, Modeling rewards and incentive mechanisms for delay tolerant networks, In 2014 12th International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOpt), (2014), 233–240.

20.

Seregina

, Brun

, El-Azouzi

and Balakrishna Prabhu

, On the design of a reward-based incentive mechanism for delay tolerant networks, IEEE Transactions on Mobile Computing 16(2) (2016), 453–465.

21.

Mao

and Zhu

, A game theoretical model for energy-aware DTN routing in MANETs with nodes’ selfishness, Mobile Networks and Applications 20(5) (2015), 593–603.

22.

Raja Meena

and Talwai

, Comparison of flat routing protocols of wireless sensor network, In International Journal of Advanced Computational Engineering and Networking 3(10) (2015).

23.

Rasbi

K.A.

, Shaker

and Sharef

, Survey on data-centric based routing protocols for wireless sensor networks, arXiv preprint arXiv:1705.04069, (2017).

24.

Ayub

and Rashid

, Resource refrain quota based routing protocol for delay Tolerant network, Wireless Networks 2(8) (2019), 469–4704.

25.

Mao

, Zhou

, Ling

and Lloret

, An optimized probabilistic delay tolerant network (DTN) routing protocol based on scheduling mechanism for internet of things (IoT), Sensors 19(2) (2019), 243.

26.

, Zhang

and Yang

, Reliable Energy-Aware Routing Protocol in Delay-Tolerant Mobile Sensor Networks, Wireless Communications and Mobile Computing 2019 (2019).

27.

Jain

and Verma

, Bubble Rap Incentive Scheme for Prevention of Node Selfishness in Delay-Tolerant Networks, In Smart Innovations in Communication and Computational Sciences (2019), 289–303.

28.

Jiang

, Men

and Tian

, A Credit-Based Congestion-Aware Incentive Scheme for DTNs, Information 7(4) (2016), 71.

29.

Mao

, Li

X.-Y.

, Song

W.-Z.

, Xu

and Moaveni-Nejad

, Energy efficient opportunistic routing in wireless networks, In Proceedings of the 12th ACM international conference on Modeling, analysis and simulation of wireless and mobile systems, (2009), 253–260.

A comprehensive approach to avoid node selfishness and data redundancy

Abstract

Keywords

1 Introduction

2 Related works

3 Network model and proposed method

3.1 Network model

3.3.1 Direct diffusion (DD)

4.1.1 Delivery ratio

Table 6 Overhead based on buffer size with existing methods Buffer size Proposed Method Spray and Wait IRS MuRIS 4 15 43 32 23 8 21 45 34 27 12 25 48 36 32 16 27 52 42 36 20 30 58 45 39

References

Table 6
Overhead based on buffer size with existing methods

Buffer size Proposed Method Spray and Wait IRS MuRIS

4 15 43 32 23

8 21 45 34 27

12 25 48 36 32

16 27 52 42 36

20 30 58 45 39