Quality-of-Service-Linked Privileged Content-Caching Mechanism for Named Data Networks

Version	Summary	Created by	Modification	Content Size	Created at	Operation
1		SHRISHA H S	--	2479	2022-06-03 05:35:10	\|
2	update references and layout	Amina Yu	Meta information modification	2479	2022-06-06 04:14:18	\|

This entry is adapted from the peer-reviewed paper 10.3390/fi14050157

The domain of information-centric networking (ICN) is expanding as more devices are becoming a part of connected technologies. New methods for serving content from a producer to a consumer are being explored, and Named Data Networking (NDN) is one of them. The NDN protocol routes the content from a producer to a consumer in a network using content names, instead of IP addresses. This facility, combined with content caching, efficiently serves content for very large networks consisting of a hybrid and ad hoc topology with both wired and wireless media.

Named Data Network (NDN) information-centric networking (ICN) quality of service (QoS)

1. Introduction

Internet technology has been transforming in the form of all-pervasive, ever-connected devices, which opens opportunities for applications and novel processes. Real-time and critical systems require high reliability and low response times in the application fields of Industry 4.0 ^[1], autopilot vehicles ^[2], disaster management, and emergency response situations. Networks often tend to use low power, lossy networks, and avail link-layer protocols such as LoRAWAN and NB-IoT ^[3], but quality of service (QoS) is not implicit, and the burden falls on the network layer to implement and achieve the desired reliability for data transmission.

The Named Data Network (NDN) ^[4] is based on an information-centric networking (ICN) paradigm, introduced as an additional protocol which complements internet protocol (IP). There are two important packets in NDN, namely, “interest” and “data”. There are three components in NDN for stateful forwarding, namely, content stores (CS) ^[4], forwarding information bases (FIBs) ^[4], and Pending Interest Tables (PITs) ^[4]. The consumer node issues interest. The successive NDN node, upon receiving a consumer interest packet, performs a lookup in its CS for the requested content; PITs perform the same lookup for the content name entry. A CS is a memory of variable size used to store any data which flows through the node temporarily. If the consumer request matches with the data available in the CS, then the interested data is served; otherwise, the content name, along with its associated incoming interface (information of neighboring node from which the interest packet was received), is recorded in the PIT. If the content name entry already exists in the PIT, indicating that there are other consumers waiting for the said content, then the incoming interface of the new interest is added to the content name entry of the PIT. The unsatisfied interest packet is forwarded to next node. An FIB maintains the interfaces of the neighboring nodes. The content names in the PIT are alive until the request is either satisfied or dropped due to expiry of time, pre-emption and so on. The sequence of interest-packet forwarding repeats until the interest reaches either the source of the content, or any CS of intermediate nodes where valid content is present. Once the content is found in an NDN node, the data packet (with requested content) traverses the reverse path by referring to the PIT entry and its associated interfaces to reach the consumers. The CS of the NDN nodes along the path may decide to cache the traversing content, depending upon their admission control strategy. The unique feature of an NDN is that every time a consumer requests content, the request need not be satisfied by the producer of that data. Intermediate nodes may have cached the data in their CS, and the data can be served directly from a CS. Content names are organized in a hierarchical manner, and are addressed as depicted in Figure 1 ^[4].

Figure 1. NDN operational scenario ^[4].

There is increasing research towards employing NDNs for communication network applications ^[5]^[6]^[7]. Research studies ^[8]^[9] conclude that NDNs can constitute reliable network-layer protocols for networks with acceptable network throughput and latency. A survey of the existing one shows that there is very little exploration on the QoS properties of NDNs from the context of content caching for networks which use a low-power network architecture, such as the Internet of Things (IOT). In an IP network, QoS involves managing network resources such as bandwidth, buffer allocation, and transmission priority ^[10]. An NDN-based network adds an additional dimension of resource management, i.e., CS management and PIT management.

The implementation of QoS for a network involves assuring resource priority to the qualified content stream over the regular content stream. Flow-based QoS guarantees the priority for a content stream based on the application of origin, and class-based QoS guarantees the priority for a content stream based on traffic classes such as telephony, network control packets and so on ^[11]. In NDNs, content may have originated from multiple sources, and may be consumed by multiple destinations. For this reason, content-forwarding resources are distributed, and these NDN properties make QoS enforcement difficult. The request–response sequence in an NDN navigates in a hop-to-hop manner, which may trigger an unforeseen amount of data packets and quickly exhaust the resources. Pre-empting pending interests, dropping data packets, and prioritizing the interests of QoS by differentiating regular traffic without a proper strategy may lead to a burst in interest packet volume, and resources may become underutilized ^[12].

For NDNs, QoS strategies can leverage the inherent advantages of a CS. A CS improves turn-around time, reduces traffic congestion by availing the content in demand through multiple NDN nodes, and acts as a re-transmission buffer. QoS strategies should consider managing PITs, which indicate pending interests to be satisfied. As priority interests become satisfied by data packets, regular interests recorded in a PIT may be starved. In another situation, a CS and PIT may become saturated with QoS traffic and become unable to cache further content and maintain the record of pending QoS content interests, respectively. In such situations, heuristic mechanisms are needed to manage the CS and the PIT with proper service balance for the regular and competing QoS content. While managing distributed resources such as PITs and CS, coordination between NDN nodes are essential. Identical content in a cluster of NDN nodes decreases the potential of the network ^[13]. The overall CS capacity of the NDN may be efficiently used if a cooperative caching scheme is implemented. In the context of PITs, a saturated PIT may pre-emptively remove entries and terminate the path for the content flow. This wastes the content-forwarding resources of preceding NDN nodes. This is unacceptable ones where the traffic is QoS guaranteed. In this light, this proposal explores ways to manage the PITs and CS of NDNs to implement QoS in NDNs.

The Internet Engineering Task Force (IETF) has published guidelines on tackling QoS issues in information-centric networks using the techniques of flow classification, PIT management, and content store management from a resource-management point of view ^[14]^[15]^[16]. The traffic flows are identified as, and in the order

for “prompt” QoS, which aims to minimize the delay for QoS traffic by providing a content-forwarding priority. For “reliable” services, the priority order is

with the intention of guaranteeing the reliable delivery of data packets to consumers. The nomenclature “prompt”, “reliable”, and their priority orders are preserved in QLPCC, as proposed by the IETF publication. The flow classification introduces the concept of embedding name components to establish equivalence among different traffic flow priorities ^[14]^[15]^[16]. The proposed QoS-linked privileged content-caching (QLPCC) mechanism employs this concept to facilitate the assignment of priorities for different traffic flows indicated by the Flow ID. A Flow ID is a name component which uniquely identifies a priority traffic flow for the implementation of flow differentiation. QLPCC proposes the calculation of an eviction score for a given priority content entry, for eviction from a saturated PIT. Content store management consists of an admission-control phase, where heuristics decide whether or not to cache the content. Each content belongs to a traffic flow, and each QoS node adopts a Flow ID. If the content belongs to the adopted Flow ID, then it is cached by the CS; otherwise, the content is forwarded without caching. The admission-control algorithm defines the steps to adopt new Flow IDs and to drop the old ones. The second phase of CS management is content eviction for the better utilization of limited available memory. QLPCC proposes a novel heuristic-based content-eviction algorithm by considering their QoS priorities, usage frequency, and content freshness. Time-expired content is evicted first, followed by prompt-priority content which was overlapped by Least Frequently Used (LFU) and Least Fresh First (LFF) algorithms. If such an overlapping was not found, then LFU prompt-priority content is evicted. If a memory requirement still persists, then the content-eviction strategy applies LFU and LFF algorithms on reliable priority content. Content that is overlapped by LFU and LFF is evicted first, followed by LFU reliable-priority content.

The proposed QLPCC designates dedicated NDN nodes as QoS nodes for implementing the proposed strategies. QoS nodes are resourceful NDN nodes that handle QoS traffic flows as per the QLPCC strategy. A QoS node can be any resourceful NDN node, including routers. The unique feature of the proposed QLPCC is that it provides support for privileged content through both PIT management and CS management.

QLPCC is simulated on an ndnSIM ^[17] platform, and results are compared with EQPR ^[18], PRR ^[19], probability cache, and LFU and LFF schemes. QLPCC outperformed the previously mentioned schemes in terms of content store hit rate, response time, and hop count reduction from the perspective of priority traffic and overall traffic. QLPCC is also evaluated in terms of content store hit rate vs. the percentage of QoS nodes in the network.

The contributions of the proposed QLPCC strategy are listed as follows:

The QLPCC strategy to manage QoS content in NDN-based networks;
A QoS-based PIT management scheme through a novel Flow Table involving the calculation of eviction scores;
A QoS-based content store management scheme, which proposes the content store admission-control and content-eviction heuristics;
A priority flow adoption method for caching QoS-linked privileged content among QoS nodes;
A reduction in hop count, an increase in content store hit rate, and an improvement in response time.

2. NDN and QoS

Research groups under the IETF have explored the issues of traffic flow classification ^[14], congestion control QoS services for ICNs ^[15], and resource management for content with different priorities ^[16].

2.1. Flow Classification

Flow classification is the foundation of QoS. Flow classification is a problem of grouping the packets and identifying the priority at which forwarding resources must be allocated for the said group ^[14].

In an IP network, flow classification is executed using a source address, destination address, source port, destination port, and protocol type identified with the packet. NDN packets cannot be identified by their source and destination addresses for flow-identification purposes. Therefore, an alternative mechanism has been proposed by the Internet Engineering Task Force (IETF);
The equivalence class name component type (ECNCT) ^[15] introduces name components which identify a particular flow and infer the equivalence classes of the traffic. This mechanism does not need any alterations in the NDN paradigm and easily integrates with the existing framework. The name component can be encoded at any granularity of the name hierarchy of the content, and this facility can be used to identify equivalence classes of streams and sub-streams of content at any desired granularity.

Consider the named content “Netflix/Show1-Prompt/Frame ID/#Segment”: all content of the streaming service Netflix may not require QoS, but a subset of the content may. An equivalence class identifier with encoded naming conventions can establish the equivalence class without any additional overhead if the naming conventions are recognized by the network stakeholders. The content consumer issues “interest packets” to the NDN network, indicating the equivalence class. The content producer/intermediate NDN node streams “data packets” indicating the corresponding equivalence class which may be served to multiple destinations, either through a direct path or through a CS in a non-synchronized way. The ECNCT does not add any additional overhead to the existing packet structure, but requires the addition of a “Flow Table”, which can identify the name prefixes for establishing class equivalences. Flow classification has useful applications for enforcing forwarding-rate control for the content of equivalent classes, the estimation of unique flows traversing through a given bottleneck, which is useful in congestion control, and to make caching decisions.

2.2. Issues to Consider for Implementing QoS Services for NDN

Congestion control in an NDN is aimed at preventing network overload, preventing the starvation of a particular class of traffic, and implementing fairness in resource allocation;
Leveraging the NDN name hierarchy is beneficial for traffic classification, rather than framing a separate definition. The network may use CS as an instrument for implementing temporary re-transmission buffers and avoiding content request load on the producers;
The resources which can be managed to implement QoS in an NDN are bandwidth, content stores, and PITs ^[15];
A content name entry into the PIT ensures sufficient bandwidth is allocated to the content in the inverse path towards the consumer, but this time-invariable interest entry may have to wait for a long time to become satisfied by the corresponding data packet, causing inefficient PIT space and bandwidth reservation;
For managing NDN resources, policies for the identification of traffic equivalence classes and their corresponding treatment must be specified;
As the consumer-requested content may be satisfied by multiple sources, the effect of topology plays an insignificant role for QoS;
QoS mechanisms of IP cannot be directly ported to NDN because hop-to-hop transmission is not confined to a single path for an NDN, thereby restricting the ability of advanced resource allocation during the process of network admission control.

2.3. Resource Management for Prioritized Content

(Prompt, reliable), prompt, reliable, and regular are the four flow priorities recognized for the implementation in QoS ^[16]. Two forwarding queues, one for prompt forwarding and another for regular forwarding, are utilized. In this light, three situations of decision making in the context of content forwarding and caching are realized:

Local cocooned decisions, which are not inter-related with any other resources;
Decisions based on locally related network resources;
Decisions based on globally distributed resources.

Local cocooned decisions do not refer to the status of other resources or mechanisms while making caching or forwarding decisions. Content is allotted to the respective queues according to their priorities. Reliable-priority content is provided prominence over regular-priority content for caching purposes. In PIT management, prompt-priority entries will replace regular-priority entries if the PIT of the NDN node is saturated. In situations where decisions are based on locally related network resources, caching and PIT management operations are based on the validity of a PIT entry and the status of the prompt forwarding queue of a given NDN node. Here, the word “local” refers to intra-device resources. If the arriving content complements a PIT entry, the said content is forwarded with reference to its priority. Prompt queue is full, the prompt-priority content will be assigned to regular queue and provided priority over regular content. Caching decisions follow the order of (prompt, reliable), reliable, prompt, and regular, respectively, while recognizing priority levels for adjusting the weights for content-caching algorithms. From the view point of globally distributed resources, the focus is on maintaining uniformity across PIT and CS management schemes in terms of QoS policies.

References

Javaid, M.; Haleem, A.; Pratap Singh, R.; Suman, R. Significance of Quality 4.0 towards comprehensive enhancement in manufacturing sector. Sens. Int. 2021, 2, 100109.
Wotawa, F.; Peischl, B.; Klück, F.; Nica, M. Quality assurance methodologies for automated driving. Elektrotech. Informationstechnik 2018, 135, 322–327.
Sallum, E.; Pereira, N.; Alves, M.; Santos, M. Improving Quality-Of-Service in LoRa Low-Power Wide-Area Networks through Optimized Radio Resource Management. J. Sens. Actuator Netw. 2020, 9, 10.
Zhang, L.; Afanasyev, A.; Burke, J.; Jacobson, V.; Claffy, K.; Crowley, P.; Papadopoulos, C.; Wang, L.; Zhang, B. Named Data Networking. SIGCOMM Comput. Commun. Rev. 2014, 44, 66–73.
Touati, H.; Aboud, A.; Hnich, B. Named Data Networking-based communication model for Internet of Things using energy aware forwarding strategy and smart sleep mode. Concurr. Comput. Pract. Exp. 2021, 34, e6584.
V, A.; Kiran, A.G. SynthNet: A skip connected depthwise separable neural network for Novel View Synthesis of solid objects. Results Eng. 2022, 13, 100383.
Shrisha, H.; Boregowda, U. An energy efficient and scalable endpoint linked green content caching for Named Data Network based Internet of Things. Results Eng. 2022, 13, 100345.
Gündoğan, C.; Kietzmann, P.; Lenders, M.; Petersen, H.; Schmidt, T.C.; Wählisch, M. NDN, CoAP, and MQTT: A Comparative Measurement Study in the IoT. In Proceedings of the 5th ACM Conference on Information-Centric Networking, ICN ’18, Boston, MA, USA, 21–23 September 2018; Association for Computing Machinery: New York, NY, USA, 2018; pp. 159–171.
Chakraborti, A.; Amin, S.O.; Azgin, A.; Misra, S.; Ravindran, R. Using ICN Slicing Framework to Build an IoT Edge Network. In Proceedings of the 5th ACM Conference on Information-Centric Networking, ICN ’18, Boston, MA, USA, 21–23 September 2018; Association for Computing Machinery: New York, NY, USA, 2018; pp. 214–215.
Karamchati, S.; Rawat, S.; Yarram, S.; Ramaguru, G.P. Mapping mechanism to enhance QoS in IP networks. In Proceedings of the 2018 International Conference on Information Networking (ICOIN), Chiang Mai, Thailand, 10–12 January 2018; pp. 797–803.
Chen, X.; Li, Z.; Zhang, Y.; Long, R.; Yu, H.; Du, X.; Guizani, M. Reinforcement learning–based QoS/QoE-aware service function chaining in software-driven 5G slices. Trans. Emerg. Telecommun. Technol. 2018, 29, e3477.
Gündoğan, C.; Pfender, J.; Kietzmann, P.; Schmidt, T.C.; Wählisch, M. On the impact of QoS management in an Information-centric Internet of Things. Comput. Commun. 2020, 154, 160–172.
Yu, Z.; Hu, J.; Min, G.; Lu, H.; Zhao, Z.; Wang, H.; Georgalas, N. Federated Learning Based Proactive Content Caching in Edge Computing. In Proceedings of the 2018 IEEE Global Communications Conference (GLOBECOM), Abu Dhabi, United Arab Emirates, 9–13 December 2018; IEEE: Piscataway, NJ, USA, 2018; pp. 1–6.
Flow Classification in Information Centric Networking, Internet-Draft—Work in Progress 05. Available online: https://tools.ietf.org/id/draft-moiseenko-icnrg-flowclass-06.html (accessed on 6 January 2022).
Oran, D.R. Considerations in the Development of a QoS Architecture for CCNx-like ICN Protocols. Internet-Draft Draft-Oran-Icnrg-Qosarch-02, Internet Engineering Task Force. 2020. Work in Progress. Available online: http://www.watersprings.org/pub/id/draft-oran-icnrg-qosarch-03.html (accessed on 2 May 2022).
Gündogan, C.; Schmidt, T.C.; Wählisch, M.; Frey, M.; Shzu-Juraschek, F.; Pfender, J.; Quality of Service for ICN in the IoT. Internet-Draft Draft-Gundogan-Icnrg-Iotqos-01, Internet Engineering Task Force. 2019. Work in Progress. Available online: https://datatracker.ietf.org/meeting/interim-2019-icnrg-04/materials/slides-interim-2019-icnrg-04-sessa-qos-for-icn-in-the-iot-00 (accessed on 2 May 2022).
Mastorakis, S.; Afanasyev, A.; Moiseenko, I.; Zhang, L. ndnSIM 2: An Updated NDN Simulator for NS-3; Technical Report NDN-0028, Revision 2; NDN: 2016. Available online: https://irl.cs.ucla.edu/data/files/techreports/ndn0028-2.pdf (accessed on 2 May 2022).
Buragohain, M.; Gudipudi, P.; Anwer, M.Z.; Nandi, S. EQPR: Enhancing QoS in Named Data Networking using Priority and RTT Driven PIT Replacement Policy. In Proceedings of the ICC 2019—2019 IEEE International Conference on Communications (ICC), Shanghai, China, 20–24 May 2019; pp. 1–7.
Buragohain, M.; Nandi, S. Quality of Service provisioning in Named Data Networking via PIT entry reservation and PIT replacement policy. Comput. Commun. 2020, 155, 166–183.

© Text is available under the terms and conditions of the Creative Commons Attribution (CC BY) license; additional terms may apply. By using this site, you agree to the Terms and Conditions and Privacy Policy.

Upload a video for this entry

Information

Subjects: Computer Science, Information Systems

Contributors MDPI registered users' name will be linked to their SciProfiles pages. To register with us, please refer to https://encyclopedia.pub/register :

SHRISHA H S

View Times: 482

Update Date: 06 Jun 2022

Table of Contents

Video Upload Options

Confirm