1 Introduction

Delay Tolerant Networks (DTN) [1] are networks where an end-to-end connection is not present between source and destination node. Because of the disconnection, messages must be forwarded in participating intermediate nodes using the “store-carry-forward” mechanism, if the source node intends to send messages to the destination node [2, 3]. For data transmission, different routing protocols are proposed for DTNs [4,5,6,7].

DTN is suitable for different environments. DTNs can be used in applications in remote areas [8], satellite communication networks, military areas, underwater, wildlife tracking and much more. DTN routing faces data loss due to the lack of end-to-end connectivity. DTNs have emerged as a promising solution to address communication challenges in extreme environments, including temporary connections, delays, and limited resources. Not only these situations can be solved by DTN but also situations of daily life in cities where devices can communicate with each other without the help of the Internet.

Today’s cities and especially overpopulated areas have high density of electronic devices with high processing capabilities and good battery capacity, however limited. This situation can be exploited by DTN for the creation of various applications for the spread of messages that do not have a special importance and can tolerate the delay (dealing with the distribution of different advertisement from the shopping centers).

Still, even in this situation efficient management of energy resources is essential to keep the network stable and as long-term as possible. There are several ways to manage energy, especially in the case of an urban area where devices can be differentiated in terms of importance. Buses, cars, trams, and pedestrians with smartphones can be part of a DTN. The most sensitive devices in terms of energy will be the mobile phones of the pedestrians which will be all the time long in motion. The management of energy is delicate since if we focus too much on energy conservation, we will have a decrease in the delivery of messages and vice versa, therefore, the optimal values must be selected to have the desired network performance. The main contribution of this paper is the proposal of an integrated energy threshold and priority forwarding strategy to improve the delivery probability.

This paper is structured as follows. In Sect. 2 are shortly presented related works. In Sect. 3, is presented the proposed strategy to improve delivery probability. In Sect. 4 are presented methodology, a description of two routing protocols used in simulations and the simulation scenarios. In Sect. 5 are presented the simulation results. Finally, in Sect. 6 are presented conclusions and future work.

2 Related Work

This section will present some related works that deal with message prioritization and energy in DTNs.

In [9] authors proposed the usage of message priority and Time to Live (TTL) change to reduce congestion in disaster scenarios. A technique for message prioritization is to create hybrid protocols that depending on the situation Spray and Wait will be used to transmit messages with the lowest probability, and Epidemic and Prophet for messages with higher priority. Another technique proposed is to dynamically decrease the TTL value of messages depending on their priority and the congestion. The proposed strategies increased the delivery latency for prioritized messages. However, this study is only for disaster scenarios and better tunning of the TTL values should be done to also improve the delivery probability.

In [10] authors proposed a technique based in message priority to manage the buffer. Different from other works where the messages are transmitted in a random way, in this work the messages with higher priority are placed at the beginning of the queue. The performance of the proposed technique is compared with other techniques throughout simulations and its advantage is presented. However, the study is limited only to a given mobility scenario and energy is not considered.

In [11] a hybrid DTN routing protocol for IoT systems is proposed. The protocol works like Epidemic when there are high priority messages, as Spray and Wait for normal priority and direct delivery for low priority messages. Depending on the distance to the base station and remaining energy, for normal priority messages, the strategy used may change between Spray and Wait and direct delivery. For low priority messages the energy level is not considered. Through simulation it is shown the advantage of the proposed protocol compared with epidemic. However, the evaluation is done for a single scenario of nodes speed and mobility model.

In [12] new routing protocols that use strategies like the available energy to achieve better energy utilization during transmission and sensing are proposed. The new strategies were evaluated, and the results showed that they can improve the utilization of residual energy. However, the evaluation is done only for a single mobility model.

However, this works did not consider the remaining node energy and when the remaining energy of a node reaches a threshold value to utilize the remaining energy for transmitting important messages or considering the priority of the messages.

In this work, different from the above-mentioned works, we propose an integrated approach that considers both techniques.

3 Proposed Forwarding Strategy

The threshold approach proposed in [12] follows a simple algorithm which limits the interaction of devices if their energy level goes below a static threshold value. The flowchart of this approach is presented in Fig. 1.

Fig. 1.
figure 1

Flowchart of energy threshold approach.

Full size image
Fig. 2.
figure 2

Flowchart of the proposed integrated energy threshold and priority forwarding approach.

Full size image

In our proposed approach we use a similar logic as in the Threshold approach but with the difference that the defined threshold value it will not be considered if the packets come from a node with priority. In a way this type of algorithm tries to evaluate the importance of messages based on the devices that are being sent. The flowchart of our proposed integrated threshold and priority forwarding approach is shown in Fig. 2.

Epidemic and Spray and Wait protocols are modified to work with node prioritization. If the message comes from a tram node, it is classified as high priority (priority 1) and this message will be transmitted even if the energy level of the node is below the threshold value.

If the message comes from other nodes (cars, buses, pedestrians) it is considered as message with normal priority (priority 0) and it cannot be forwarded if the energy level of the node is below the threshold value. In the proposed strategy, if the energy level of a node is below the threshold value, and the sender priority is zero or if the energy of a node is completely depleted, the message reception is denied due to the low resources. The policy of the available buffer space is checked to determine whether there is enough space to accommodate the incoming message. After the TTL expires, the messages are dropped, and the energy level is updated.

4 Methodology

Simulation environment: The simulations were conducted in Windows 10 and the network simulator used is the-one-1.6.0 [13]. Mobility model: In these simulations it is used the model of a typical working day scenario to simulate a real environment with different mobility models like Shortest Path Map Based Movement for pedestrians and Map Route Movement for trams and cars.

A dimensioned plan 4500 m \(\times \) 3400 m is used in this simulation and the default version with the map of Helsinki with roads for cars, pedestrians and roads dedicated to trams. Other simulation parameters are presented in Table 1. The simulation time was 12 h. The main objective of this work is the comparison of delivery probability for the scenario with threshold energy model proposed in [12] and our proposed integrated approach threshold with priority forwarding. The initial energy of the nodes is the same. The energy parameters are shown in Table 2. All these comparisons will be done for the Epidemic and Spray and Wait routing protocols.

Epidemic Routing Protocol  

One of the most widely used and simple DTN protocols is the Epidemic protocol [4]. This protocol is based on flooding technique. The objective of this protocol is to maximize message delivery and minimize delays. In this protocol, nodes send copies of messages to all neighboring. They store the message in their own buffer, until they meet another node or the destination. Whenever nodes meet each other, they exchange the information in their buffer with each other comparing content and identifying new messages which must be stored on the node that does not have them. This protocol ensures an elevated level of message delivery but needs a large buffer size that saves many messages.

Table 1. Simulation parameters.
Full size table

Spray and Wait Routing Protocol  

The objective of this routing protocol is to reduce consumption of network resources [5]. Spray and Wait is similar with Epidemic protocol for information propagation speed and with Direct Delivery for simplicity. This protocol consists of two main phases. Spray: The source node produces n number of copies to transmit the message to neighboring nodes. Wait: Intermediate neighboring nodes keep the messages received in the buffer and wait until one of the nodes encounters the destination and passes the message directly.

Table 2. Energy parameters.
Full size table
Fig. 3.
figure 3

Results of Delivery Probability for Epidemic using threshold approach and the integrated threshold and priority approach.

Full size image
Fig. 4.
figure 4

Results of Average Remaining Energy for Epidemic using threshold approach and the integrated threshold and priority approach.

Full size image
Fig. 5.
figure 5

Results of Delivery Probability for Spray and Wait using threshold approach and the integrated threshold and priority approach.

Full size image
Fig. 6.
figure 6

Results of Average Remaining Energy for Spray and Wait using threshold approach and the integrated threshold and priority approach.

Full size image

5 Simulation Results

Results of Delivery Probability and average remaining energy for Epidemic protocol are presented in Fig. 3 and Fig. 4, respectively. We conducted simulations for different values of threshold. Delivery probability is decreased when the threshold value is increased because nodes will be in the active mode for less time. For Epidemic, because of the high number of copies in the network, the delivery probability is improved with the usage of priority, but the average remaining energy is decreased.

In Fig. 5 are shown the results of Delivery Probability for Spray and Wait using threshold approach and the integrated threshold and priority approach. From the results of the threshold approach, the delivery probability is decreased for higher values of threshold because the node will not be used to transmit data if the energy level reaches the threshold value. Even for the integrated approach using priority, the delivery probability decreases for higher threshold values, however since in this approach messages with priority will still be transmitted even if the threshold value is reached. The approach that uses priority has higher delivery probability compared to the simple threshold approach.

The results of Average Remaining Energy for Spray and Wait protocol are presented in Fig. 6. The increase in the threshold value leads to higher average remaining energy. For both approaches the average remaining energy is almost the same.

From the results, the proposed approach improves the delivery probability of Spray and Wait and Epidemic protocols.

The proposed approach works better with Spray and Wait protocol because the delivery probability is improved, and the average remaining energy of nodes remains almost the same.

6 Conclusions

A problem Delay Tolerant Networks (DTNs) have is energy that is usually limited and requires specific protocols and strategies for energy conservation. Setting a threshold significantly affects the remaining energy at the end of the simulation but also has a negative side as reducing the active time of connections which should be considered. The reduction of this time directly affects the lower delivery probability of packets. This can be improved by setting priorities and allowing nodes to distribute packets for specific messages that are important.

In this paper an integrated approach that combines threshold and message priority is implemented for Epidemic and Spray and Wait protocols and compared with another approach. Delivery probability is improved for both protocols when the proposed approach is used. The best results for the proposed approach are when it is combined with Spray and Wait protocol. The delivery probability is improved, and the average remaining energy of nodes remains the same.

In the future, we would like to further improve our proposed approach and evaluate the performance for different scenarios and parameters.