In particular, NDN over link layer networks will reduce the overhead of TCP/IP and improve the efficiency of data distribution. However, several unresolved issues in NDN about the link layer networks still damage NDN networks, such as broadcast overhead and MTU mismatch. In this thesis, we therefore designed and implemented an NDN Neighborhood Discovery Protocol, called NDN-NDP, to enable a unicast data transmission over the link layer.
Through CORE emulation, we have demonstrated that our NDN-NDP outperforms previously proposed NDN link layer technologies in terms of transmission overhead, network throughput, delay, and a number of unsatisfied interests. Our NDN-NDP can reduce transmission costs, delays, and a number of unsatisfied interests by approximately 99%, 46%, and 11%, respectively.
INTRODUCTION
A unicast transmission mode in NDN over a link-layer network has been provided by developing NDN-NDP protocol. NFD has been extended to enable NDN-NDP protocol and aMTU for NDN over the link-layer network. While NDN supports many transport services, this work is only focused on NDN over a link-layer network.
The goal of this thesis is to solve the broadcast overhead and MTU mismatch problem for NDN over the link-layer networks. It is followed by NDN concepts and mechanisms, NDN forwarding process and NDN over the link layer network.
BACKGROUND AND RELATED WORK
- Networking Trends
- The Connected Machines
- Evolution of the Internet
- Future Internet Architecture
- Critical Problems in Modern Networking
- Inefficient protocol functionality in mobility networks
- Inflexible protocol architecture
- Lack of security
- Content Distribution Problems
- Named-Data Networking
- Named Forwarding Daemon
- NFD Modules
- Face System
- MTU Mismatch Problems
- Previous Solutions
The IP home glass, as shown in Figure 1, represents an elegant model of point-to-point communication. 15] have shown that HTTP is a bottleneck of the Internet, as depicted in the application layer of Figure 1. While applications and network hardware are still evolving, IP is still a bottleneck of interest.
Unlike routing table freshness, stateless is a notorious problem when using IP. Since MANET was developed, research topics on MANET have expanded to various types of specific applications [40–42] (such as the Vehicular Ad-Hoc Network (VANET)). For NDN on the right of Figure 3, a content part plays an important role in transmitting NDN data.
If the content exists, the NFD immediately forwards data packets of the desired content from the CS to the consumer. On the reverse path back to the consumer, data packets may be cached in the CS of intermediate nodes. Therefore, MTU mismatch is one of the important problems for NDN in link layer networks.
As shown in Figure 7, NDNLP replaced the IP layer and worked on top of the link layer. For the index of the last fragment, B starts the process of reassembling the frame packets and forwards the data packet to the NFD. Finally, intermediate reassembly and interrupted transmission of NDN fragments were considered to address NDN fragments.
A data packet name was used to design a forwarding path in highly dynamic wireless networks. A physical mobility of the network node and a logical mobility of data can be managed.
RESEARCH METHODOLOGY
Introduction
An analysis of the analytical modeling shows that it is not suitable for studying our NDN-NDP. Computer networking researchers generally accept network simulation as an important tool to explore new networking paradigms. To evaluate through network simulation, discrete events in the network simulation help researchers understand protocol behavior.
Experimental results on the testbed network can give a high accuracy because the right hardware is used. In Figure 12 and Figure 13, for example, an NDN research community has collaborated to provide a global NDN testbed. In this dissertation, simulation tools were investigated to select as a tool to experiment NDN-NDP.
To validate the CORE emulator, a simple network topology was created, as shown in Figure 18. There are three nodes in this simple topology, including a consumer, a producer, and an NDN forwarder. Through a validation process, experimental results of this validation must be valid and robust for a statistical analysis.
Nevertheless, the CORE emulator results are still stable and reliable for use in this dissertation. The experimental results of NDN-NDP and NDNLP would be generalized and produce sufficient quantity to make an accurate decision. To obtain a reliable result, the mean value of the experimental results is marked with a confidence interval of 95%.
OUR PROTOCOL DESIGN
Introduction
As shown in Figure 20, an overview of the network topology with an NDN-NDP table. NDN-NDP helps create an NLUF and produces the NDN-NDP table for performing the unicast transmission. In NDNLP, NDN Link-layer Broadcast Face (NLBF) is the only choice for communication over NDN link-layer networks [4].
Therefore, we propose a novel NDN Link-layer Unicast Face (NLUF) that supports unicast transmission over link-layer networks. To support unicast communication over link-layer networks, we need to map a destination named prefix to a destination unicast MAC address. This NLUF maps a unicast MAC destination address to a local face at the link layer in the Face Table (as shown in Figure 21).
In the FIB (shown in Figure 23), this NLUF is further mapped to a destination prefix named “/lab/isan/movie1”. NDN-NDP is installed to learn target MAC addresses and map target named prefix to target MAC addresses. The "name" field of this NDI consists of the NDN-NDP identifier, the device MAC address, the interface MTU, and an optional named prefix (as shown in step 1 of Figure 24).
If node A is a producer, the prefix will indicate the named prefix of node A, for example "prefix=/ndn/uk/ac/leeds/nodeA". NDI from node A will reach all other NDN nodes in its link layer network. If a named prefix (in this example, "/") is attached to the NDI, Node B will map the NLUF to the named prefix in the FIB.
Adaptive MTU
The NRD must also be digitally signed to be validated in the next step. After receiving the NRD from node B and verifying its legitimacy, node A establishes a new NLUF in node B, as shown in step 3 of Figure 24. After the NLUFs between node A and B are successfully established, NDN communication between these two nodes over the link layer network can be made in unicast mode.
An inactive NLUF for a period of time will be destroyed provided the link has been broken. To keep NLUFs alive, both A and B maintain their NLUFs by sending a special interest message (called the "NLUF maintenance message") to each other every Th. Without receiving the NLUF maintenance notice for a specified period of time, the NLUF record is deleted.
To protect against similar problem, NDN over link layer networks needs a verification of interest and data packets before making a unicast transmission. Signature: The digital signature on an NRD packet must be verified before any action is taken. It is not in the scope of this dissertation, but should be done as future work.
To complete our NDN operations described earlier, NDI inserts /ndp as the first named prefix of the NDI packet to identify the NDN-NDP protocol. An optional named prefix in the special parameters section will be inserted if the NDI sender can act as an NDN forwarder. For the last field, the Nonce is used to verify with the incoming NRD, which is required to match the NDI packet.
PERFORMANCE EVALUATION
The number of broadcast packets: Broadcast packets in an NDN link layer network are actually the multicast packets, which send over the default ICN-MCAST (MAC address 01:00:5E:00:17:AA). The number of unsatisfied interests: The number of unsatisfied interests can be used to measure forwarding performance. In our experiments, the number of broadcast interest packets and the number of broadcast data packets between NDN-NDP and NDNLP were compared for three different MTU settings.
In summary, by proposing unicast faces, NDN-NDP can dramatically reduce the number of transmission packets compared to NDNLP. This success is because the unicast faces of NDN-NDP use the link layer network more efficiently than the multicast faces of NDNLP. The comparative delay between NDN-NDP and NDNLP for different MTU settings is illustrated in Figure 29.
Therefore, the overall experimental results indicated that NDN-NDP can reduce delay, which effectively improves NDN over link-layer networks. From the experiments, the number of unsatisfied interests of NDN-NDP and NDNLP were counted comparatively. The total number of unsatisfied interests of the NDN-NDP is therefore smaller than the number of unsatisfied interests of the NDNLP.
NDNLP suffers from both broadcast overhead and MTU mismatch problems, while our NDN-NDP is designed to avoid the MTU mismatch and reduce the broadcast overhead. However, BEF is based on the CCNx forwarder, while our NDN-NDP uses an NDN forwarder. However, we argue that µMTU of FIGOA would be inefficient compared to aMTU of our NDN-NDP.
CONCLUSIONS AND FUTURE WORK
The experimental results illustrated that: 1) NDN-NDP can cut the number of broadcast interest packets and the number of broadcast data packets by about 99%;. This thesis evaluated the previous proposals of NDN over link layer networks, and successfully proposed a new design of NDN-NDP. We analyzed NDN over the link layer network, we found performance criteria to evaluate our NDN-NDP.
Our NDN-NDP is a protocol to map between MAC addresses and named data to enable a unicast transmission mode. For example, our NDN-NDP can be used in other network areas (such as IoT and MANET) to reduce broadcast overhead, delay, and the number of unsatisfied interests. Heartbeat interval: In our NDN-NDP we have placed a heartbeat interval in our NDN-NDP protocol.
Raychaudhuri, "MobilityFirst Future Internet Architecture Project," i Proceedings of Asian Internet Engineering Conference, Bangkok, Thailand, 9. - 11. november 2011. Tschudin, "An Information Centric Network for Computing the Distribution of Computations," i Proceedings of ACM Conference on Information -Centric Networking, Paris, Frankrig, 24. - 26. september 2014, s. Wählisch, "The Need for a Name to MAC Address Mapping in NDN: Towards Quantifying the Resource Gain," i Proceedings of ACM Conference on Information-Centric Networking, New York, NY, USA, 2017, s.
Zhang, "On Broadcast-based Self-Learning in Named Data Networking", in Proceedings of IFIP Networking Conference, Stockholm, Zweden, 12-16 juni 2017. Zhang, "NDN-NIC: Name-based Filtering on Network Interface Card", in Proceedings of ACM Conference on Information-Centric Networking, Kyoto, Japan, 26 - 28 september 2016, pp. Festor, "Server-side Performance Evaluation of NDN", in Proceedings of ACM Conference on Information-Centric Networking, Kyoto, Japan, 26-28 september 2016, blz.
![Figure 11 An event-based process in simulator Source: [60]](https://thumb-ap.123doks.com/thumbv2/filepdfco/10231020.0/30.892.166.780.184.642/figure-11-event-based-process-simulator-source-60.webp)
