38
the competition of network storage resources and easily causes network congestion.
39
Figure 2-7 Example of a ferry node.
The ferry node is introduced to assist other nodes in message delivery and improve network connectivity. In order to achieve its goal, the ferry node needs to have the following
characteristics:
1) Controllable movement model: The ferry node is a kind of node that the movement model is controllable. Through a reasonable configuration of parameters such as the movement path of the node, it can play its role to the maximum extent, deliver messages for other nodes, and make the divided network reconnect [134].
There are two kinds of nodes with a controllable movement model: one is deployed for non-communication reasons. For example, the primary function of a bus is to provide public transport services. It is not specifically set for communication, but as it moves on a certain route, it can carry messages and act as a ferry node [135]. Another category is designed specifically for communication, for example, when a disaster happens, some UAVs or other equipment can be deployed to shuttle between high-risk areas and other areas where communication is difficult to provide transferring messages in those areas [136].
40
2) Repeatable movement: If the movement of nodes is not repeatable, it can only visit other nodes once, although it can forward messages to other nodes, it cannot play an obvious role in network connectivity. When a ferry node is deployed, it is often necessary to repeatedly visit the same area to improve its ferry performance. It should be noted that the repeatability requirements of ferry nodes are not strict, and the order and time of visiting other nodes can be changed [137].
3) Large buffer size: The purpose of deploying a ferry node is to deliver messages to other nodes. Therefore, compared to a normal node, a ferry node must have a large buffer size to save the received messages during the running cycle and forward them when they encounter the destination node. If the buffer size is insufficient, data will be lost, and the role of the ferry node will be affected [138].
Although ferry nodes have a large buffer space, if the running cycle of ferry nodes is long, and the cache will still overflow. Therefore, when designing the movement model or routing protocol of the ferry node, it is necessary to consider the abundant energy resources [139].
4) Abundant energy resources: The movement of the ferry node needs to consume more energy, so in order to ensure its continuous operation, it needs to be equipped with a larger capacity of batteries. In practical use, the ferry node is usually large equipment such as UAVs and vehicles, so the energy consumption of the ferry node is generally not considered [140].
2.6.2 UAV technologies in VDTN
UAVs can receive and forward data as a ferry node in the air, improving the reliability of wireless communication links, enhancing infrastructure flexibility, and providing communication coverage in sparse areas [141] [142]. UAVs are mainly used in military fields for joint operations [143]. In recent years, UAVs have been gradually applied to civil industries such as intelligent transportation systems [144], earthquake relief [145], and precision agriculture [146], especially the rapid development of IoT and the 5G industry [147]
[148], which have set off a wave.
Based on the traditional message ferry algorithm, researchers study the routing for
41
multipath ferry under the condition that the number of ferry nodes is limited, and four algorithms were proposed: single routing algorithm (SIRA), multi-routing algorithm (MURA), node relay forwarding algorithm (NRA) and ferry relay forwarding algorithm (FRA) [149].
The SIRA algorithm optimizes the UAV flying path when all UAVs follow the same path.
The MURA algorithm optimizes multiple flying paths for UAVs. In addition to multiple flying paths, the NRA algorithm uses some preinstalled fixed network nodes to provide information exchange between different UAVs belonging to different flying paths. The FRA algorithm optimizes multiple UAV flying paths by considering the encounter probability between different UAVs from different flying paths but without the pre-installed fixed network nodes.
In the Optimization Way Point ferry routing (OPWP) [150], the movement path of a single ferry node is determined by calculating the probability between the ferry node and other nodes according to a specific movement model.
Burns and Brock studied the control of autonomous agents to improve the performance of DTN, which is feasible for arbitrary movement of nodes. This study does not disrupt the mobility of nodes [151].
Other related work includes the study of integrated caching and path management schemes in DTN with ferry nodes. In this work, a cache allocation scheme is designed for message ferries using the max-min fairness model [152].
These works reflect the achievements of researchers in the field of DTN routing and the progress of message ferry technology in the application of DTNs. However, due to the harshness of the DTN environment, the current routing design focuses on either saving storage and energy consumption, reducing transmission delay, or successful data transmission rate, so it is difficult to take both into account [153].
Through a period of study and accumulation, UAV assisted VDTN technology has made great progress. However, there are still some key problems that have not been well solved.
The UAV assisted VDTN is faced with the challenge of network interruptions or intermittent connectivity. There are still no unified standards and methods for network modeling, connectivity definition, and measurement. Connectivity in UAV assisted VDTN is based on two dimensions of time and space. The current research is still limited to networks that are
42
fully measurable in advance for node movement and contact opportunities in the future, which also limits the application scope of UAV assisted VDTN.
When developing and designing the UAV assisted VDTN routing protocol, defining system performance in the UAV assisted VDTN is still an open question. When circumstances permit, routing protocols should be designed with node locations, topologies, and future trajectories in the network.