Mohamed Amine Korteby1 – Zoltán Gál2 – Ashraf Dabbas3
1 Ph.D. Student , University of Debrecen (Faculty of Informatics), Korteby.Amine@inf.unideb.hu;
2 Ph.D., University of Debrecen (Faculty of Informatics), Gal.Zoltan@inf.unideb.hu;
3 Ph.D. Student , University of Debrecen (Faculty of Informatics), Ashraf.Dabbas@inf.unideb.hu
Introduction
Wireless sensor network is one of the structures being used for exchanging data rapidly and correctly (Meng S. 2011). Capable to be deployed with low coast in a harsh environment, where replacing or renewing their power supply becomes difficult or impossible (Kurt S. 2017). They are composed of a set of nodes that can be mobile or static, communicating via short-range wireless protocols, sensing diverse phenomena and forward the information to the destination (Chen C.P. 2015).
Other types of networks are Delay Tolerant Networks (DTNs) or Opportunistic Networks (ONs) that advanced directly from the Mobile Ad-hoc Networks (MANETs), lack of infrastructure and limited resources are the common aspects between the networks. However, they come with several diversities. MANETs sensors interconnect using basic TCP/IP protocol, while DTNs use Bundle protocols, invent to support the store to carry transmission standard to communicate with nodes (Sobin CC. 2016).
Emerging applications usually comprise challenging computations and the need for real-time requirements such as bounded end-to-end delay and guaranteed Quality of Service. Divers application can be found on WSN such as intrusion detection, weather monitoring, disaster management and detecting ambient, every one of them has a specific task to be done and depending on the data captured, some applications need immediate action to be taken after collecting event to transmit the sensed data must to the destination in a short time (Bamatraf A. 2015).
Abstract: Several Wireless Sensor Network (WSN) applications function autonomously in defective or unreachable environments, preventing maintenance or human intervention. Redundant deployment is usually considered in this scenario, making the network robust to failure and environmental changes.
New applications need a connection of end-to-end delay and high reliability for various communication models. In current WSNs, the QoS deteriorates upon augmented traffic load due to congestion, the shared medium, large number of hops in data routing and the overall coverage area issue. This paper investigates the impact of the Shortest Map-Based movement model on the quality of service (QoS) of the sensor network, under three routing protocols with nine Time-To-Live (TTL) values for a simulation of ten hours each.
In this paper, we analyze the behavior of three routing protocol under the Geographic Map-Based mobility models and parameters, using the Opportunistic Networking Environment (ONE) simulator which is an extensible tool for evaluating protocols and mobility models for WSN. We use the provided data to analyze the impact of the movement model on the communication quality, delay and packets loss of the network by changing the TTL value and the population density of the WSN.
The rest of the paper is organized as follows. Related work is discussed in section 2. General description of the used routing protocols and mobility model is given in section 3. The measurement scenarios and the behavior description of the WSN in different cases is given in section 4. Finally, conclusions and possible perspectives of this work are presented in section 5.
Related work
S. Pramono and al compared the performance of the ZigBee protocol in two topologies (star and multihop) according to parameters quality of service, delay, throughput, and packet loss. They also measured the signal strength is for each node placement with RSSI analysis to prove whether it is related to the quality of service results obtained. A ZigBee protocol was used that supports both topologies and a circuit equipped with MQ-9 sensor which will generate sensor data and then sent from the end device node to the coordinator node (Subuh P. 2017).
G. Horvat, J. Vlaović and D. Žagar made an experimental validation for query-based communication in a star topology for various numbers of nodes and packet generation interval is performed with the proposition of a cross layered handover algorithm to improve deteriorated performance and the QoS by means of multi-channel redundancy. The results show that the proposed algorithm improves the delivery rate by 25% and the RTT by 40% in the worst-case scenario. However, it underperforms upon the heavy traffic load (Goran H. 2015).
G. Horvat, D. Žagar and D. Vinko analyzed the node deployment parameters in favors of the QoS metrics by estimating critical deployment parameters and presenting a direction towards optimal solution. Since node deployment validation is not a trivial task, they approached this problem by using a set of designed experiment to obtain optimal deployment parameters in respect to QoS metrics. The parameters are modeled and optimized using Response Surface Methodology (RSM) applied to node deployment parameters. From the proposed approach results it can be concluded that RSM can give optimal parameters for node deployment in maximizing QoS by using only a small set of experiments (Goran H. 2014).
A. Bamatraf and al, classified the type of packets transmitted based on timeness and reliability. The data traffic was classified into four categories:
• Critical (C): real time and reliable packets that require both of low delay and reliable services.
• Delay sensitive (DS): real time that require low delay.
• Reliable (R): non-real time that require high reliability.
• Normal (N): non-real time that require the best effort.
The proposed protocol assures the QoS by differentiating the data type. In order to do so, the protocol assigns different priorities to satisfies the network necessities (Bamatraf A. 2017).
Characteristics of the routing protocols in mobile networks
Epidemic Routing Protocol (ERP): A node will broadcast packets to all its neighbors that are in its sensing range. Any node receiving the message will forward to its neighbors. With this technique, the messages will reach all the nodes including the destination. Summary vector method is used by the nodes for remembering each other’s to avoid broadcast storming. This protocol is ideal in the delivery time prospect in case of enough buffer size and communication bandwidth in the network.
PRoPHET (PRO): Based on probabilistic routing protocol, it uses stored information of the conveyance between the nodes to select the next-hop. This aptitude depends on spreading information in the entire network about the node’s movements with time stamps. The higher is the encounter rate of two nodes, the higher is the probability to choose the same hop by these nodes.
Direct Delivery Routing (DDR): The source sends the message only when direct communication possibility exists to the destination. No intermediate nodes are used to transfer the messages between the source and destination. It can be deduced easily that minimum network resources are used by the nodes, but the message delivery rate is very low because the probability of meeting directly the source and the destination is infinitesimal. This routing protocol type does not guarantee communication for sparsely populated network or slowly moving nodes.
Measurements and analysis of the sampled data sets
We analyzed the data captured from a WSN system. Three different routing protocols (Epidemic (R = 1), PRoPHET (R = 2), Direct Delivery (R = 3)) and one movement models (Shortest Path Map-Based) were applied in a Wireless Sensor Network, using the ONE simulator, with N (N = 128, 256, 512) nodes and nine different Time-To-Live (TTL) values. The simulation was executed eighty-one times (81 = 3x1x3x9). In each of the eighty-one simulation cases there were extracted following variables from the WSN system: remaining energy (E [%]) for each node, generated message creation time (CT), message size (MS), contact time duration (CD) of the nodes, delivered messages (D), number of hops (H) for each message and the delivery time (DT) for each message.
Description of the analyzed Shortest-Path Map Based (SPMB) movement model
Shortest Path Map-Based Movement (SPMM) Model offers basic movement model for Vehicular Delay-Tolerant Network (VDTN), where vehicles are
A) Helsinki map (Google Map) B) Helsinki road map (Simulator) Figure 1. Helsinki road map
opportunistically exploited to offer a message conveying service. This movement model uses a map of a road and Dijkstra's Shortest Path First algorithm to create a shortest path between two nodes and a points of interest (POI). The destination is reached through multiple selection of intermediate points of interest, where the shortest path is estimated. At each intermediate POI, a new path to the next POI is calculated with the same algorithm. Two consecutively destinations usually are more than one hop far from each other (Gál Z. 2018). Figure 1, represent the map of the city of Helsinki that we used in our simulation for the different routing protocols and the SPMM movement model.
Data analysis
We analyzed two aspects of the message creation, creation time difference and ECDF of message size. Figure 3, shows a uniformity of message generation at 30
Figure 2. Characteristics of the Shortest Path Map-Based model (M = 3) Left: Movement of 10 mobile terminals (nodes) in the simulated area;
Right: Movement speeds of the first 10 mobile terminals (nodes)
seconds and a linearity of message size. We noticed that for all the simulation cases and independently of the routing type (R), the number of nodes (N) and TTL values, the same results were captured by the simulator.
Figure 4 shows the bihistogram of the message generation, we can observe a uniform random distribution of creation time moments and message size. This phenomenon is caused by the input data at the beginning of the simulation. It was found that for all simulation cases, the same input sequence having size in between (500 kilobyte – 1 megabyte) is fixed and conform to the event generation rules, independently of the routing type (R), the number of nodes (N) and the time to live (TTL) parameters.
The log-log plot of the contact time duration (CD) histogram for three different densities of the population can be seen on Figure 5. The approximately linear character of these decreasing curves means that the histogram with linear scales are exponentially decaying functions. For high values of the contact time duration the histogram becomes randomly. The higher is the number of nodes, the higher is the
Figure 4. Bihistogram of the message generation:
Creation time moment (CT), Message Size (MS) Figure 3. Message generation:
Creation time differences (left); ECDF of the Message Size (right)
Figure 5. Log-log plot of the histogram of the Contact Time Duration (CD)
exponentially histogram. It can be observed that log(H256) = [log(H128) + log(H512)]/2.
This implies a special relation: (H256)2 = H128 · H512. This means that the histogram of the CD is a logarithm function of the number of nodes: HN = a·log(N), where
“a” is constant. Should be noted also that this histogram has following complex dependence: HN = a·log(N)·eCD.
The coefficient of variance of the delivery time plot, can be seen if Figure 6, where CVDT = σDT/μDT, where “σ” and “μ” are the standard deviation and mean, respectively. For each routing type (R) and time to live (TTL) parameters, we can state the following: The absolute minimum which is the lowest point over the entire domain of the function has a light dependence on the number of nodes (N), a dependence on the routing type (R) and a strong dependence on the time to live (TTL) parameters. The slightly increasing delivery time for large TTL values is due to the fact that, a large TTL means that the message can be stored in a given node for a longer time, this can decrease the priority of the message that will be sent later on to the destination.
Figure 6. Coefficient of variance (CV) of the Delivery Time (DT)
Figure 7. Coefficient of variance (CV) of the Hop Count (HC)
Figure 7 shows the coefficient of variance of the hop count, where CVHP
= σHP/μHP. Both parameters σ and μ are lightly dependent on the number of nodes (N), strongly and weakly dependent on the TTL values and the routing types (R).
As mentioned before, large TTL values mean larger the time for a given message to be stored in a node, which will give the message the possibility to traverse the network from node to node, increasing the number of hops, without the need to be delivered to the destination until a low value of TTL is reached. In cases of Epidemic (R = 1) and PRoPhet (R = 2) routing types the absolute minimum of the curves for TTL = 16 minutes shows peak quality of the WSN transmission service. With the Direct Delivery Routing (R = 3), we can observe a constant zero values. No hops were needed to send the message from source to destination, this is due to the special characteristic of the DDR, the source sends the message only when direct communication possibility exists to the destination. No intermediate nodes are used to transfer the messages between the source and destination.
Conclusion
In this paper we analyzed the Impact of the geographic map-based movement on the communication quality of a wireless sensor networks for three routing types, three node numbers and nine values of the time to live parameter. We extracted the fundamental information about the behavior of the WSN system. We stated that the same input sequence is fixed and conform to the event generation rules and only with to different routing type and other configurable parameter, the output differs. The delivery time and the hop count and not dependent on the nodes number, but strongly depended on the TTL values. Future works may continue in the context of more analysis of the map-based movement impact may be done in the next research work.
References
Bamatraf, A. (2015): Review of Quality of Service in Routing Protocols for Wireless Sensor Networks. Journal of Theoretical and Applied Information Technology, 74(3).
Bamatraf, A. (2017): Priority based QoS Protocol for Heterogeneous Traffic Routing in WSN. 6th IEEE Student Project Conference (ICT-ISPC), Skudai, Malaysia.
Chen, C.P. (2015): Efficient coverage and connectivity preservation with load balance for wireless sensor networks. IEEE Sensors J., 15(1), pp. 48–62.
Gál Z. (2018): Impact of the delay tolerance in wireless sensor networks: Buffer occupancy and energy consumption aspects. Future IoT Technologies (Future IoT), 2018 IEEE International Conference, Eger, Hungary.
Goran, H. (2014): Influence of Node Deployment Parameters on QoS in Large-Scale WSN.
3rd Mediterranean Conference on Embedded Computing MECO, Budva, Montenegro.
Goran, H. (2015): Improving QoS in Query-Driven WSN Using a Cross-Layered Handover.
4th Mediterranean Conference on Embedded Computing MECO, Budva, Montenegro.
Kurt, S. (2017): Packet Size Optimization in Wireless Sensor Networks for Smart Grid Applications. IEEE Transactions on Industrial Electronics, 64(3).
Meng, S. (2011): Zigbee Wireless Sensor Networks and Their Applications. Department of Computer Science, National Chiao Tung University, Taiwan.
Sobin, CC. (2016): A survey of routing and data dissemination in Delay Tolerant Networks.
Journal of Network and Computer Applications, 67, pp. 2483–2495.
Subuh, P. (2017): Comparative Analysis of Star Topology and Multihop Topology Outdoor Propagation Based on Quality of Service (QoS) of Wireless Sensor Network (WSN). IEEE International Conference on Communication, Networks and Satellite (Comnetsat).