Named Data Networking: A survey

Abstract

Internet was developed as a packet data network where users and data sources (server) with specific IP addresses interacted over a pre-established communication channel. This model of client–server data communication has evolved into a peer-to-peer mode of data sharing in recent times. Applications like, YouTube, Bit Torrent, social networks have revolutionized the idea of user generated contents. Modern users care only for specific data items irrespective of their sources. So, the idea of using IP addresses to identify servers hosting a particular content is losing its importance. Moreover, want of IP addresses is a challenging issue haunting the Internet community since long. The need of the time is a content-centric networking platform where data hosts are of less importance, and Named Data Networking (NDN) has been proposed to that end. NDN allows users to float a data request without any knowledge about the hosting entity. NDN can handle user mobility, security issues more efficiently than the current Internet. Although NDN has been proposed in 2010, so far, there is no survey paper studying its architecture and various schemes proposed for its different characteristic features, like, naming, adaptive forwarding and routing, caching, security, mobility, etc. In this paper, we introduce a novel taxonomy to study NDN features in depth. We have also covered several NDN applications. We conclude our survey by identifying a set of open challenges which should be addressed by researchers in due course.

Introduction

Internet was designed more than thirty years ago as a point-to-point conversation between two end hosts which allowed the users to fetch data from well-known servers. After TCP/IP protocol stack was introduced  [1], packet switching allowed users to transfer text, audio, and video packets over the Internet.

Though, the Internet has shown great resilience over the years, more recently, changes in the nature of applications, user requirements, and usage patterns have significantly strained it. Recently evolving content-centric applications, like, social networking, e-commerce, YouTube  [2], Netflix  [3], Amazon  [4], iTunes  [5], etc., allow users to share texts, images, audios, and videos and have become the source of half of the world’s Internet traffic. Recent surge on production and consumption of user generated contents (UGCs) are failing the Internet because it was not designed to support the newly evolving content distribution model  [6], [7], [8]. Today, most of the application data delivery model is concerned about what data is needed irrespective of their locations. Moreover, support for mobility and security is not in-built in Internet, but offered as multiple patches or add-on features which may fail at times.

The aforementioned reasons urged researchers to find an efficient alternative architecture to the Internet, which will inherently support content-centric communication. Among several funded projects for designing content-based future Internet paradigms, Named Data Networking (NDN) came up as the promising candidate  [6], [7], [8], [9] which directly deals with application generated variable-length, location-independent names to search and pull contents for a requesting user, irrespective of their hosting entity.

The basic design principles of NDN are based on the Internet. NDN can directly use major IP services like, Domain Name Service (DNS) and inter-domain routing policies. IP routing protocols like, BGP and OSPF can be adapted to NDN with little modifications. However, NDN offers certain enhanced features as explained below. It uses data packets with content names  [6] instead of source and destination addresses. The use of unique content names for communication allows routers to keep track of packets’ states, which supports numerous functions unlike the IP routers. The data packets are self-contained and independent from where they are retrieved and where they can be forwarded. These features allow in-network caching of contents for fulfilling future requests and inherently support consumer mobility. In NDN, all data packets are signed by its producer and verified by the consumer, unlike IP. NDN routers support multi-path forwarding, i.e., they can forward a user request to multiple interfaces at the same time. Moreover, the use of content name for communication removes the need of application-specific middleware too.

NDN and Internet share the same layered hourglass architecture with functional differences between corresponding layers  [6], [10], as shown in Fig. 1(a) and (b). The OSI communication model has only Internet Protocol (IP) in the Network layer. However, it is difficult to add new functionalities to the IP and to modify the existing ones. As a future Internet paradigm, NDNs network layer  [6], [7] must support scalability (support to large Internet topology and high amount of name prefixes), security (integrity, origin authentication and relevance of routing information), resiliency (to detect and recover from packet delivery performance), and efficiency (support multi-path forwarding and in-network caching for efficient data dissemination).

As shown in Fig. 1(b), security and strategy are two new layers added to the NDN protocol stack. Security layer provides security to each and every piece of content, unlike securing the entire communication channel in Internet. Strategy layer is used for the stateful NDN forwarding plane, which makes forwarding decision for each incoming content request. NDN does not maintain a separate transport layer. All the functions of Internet’s transport layer are embedded into the NDN forwarding plane.

NDN has evolved from the Information Centric Networking (ICN) research area as shown in Fig. 2, which has also inspired many other Internet architectures  [11]. Recently, researchers have analyzed the key features and issues of NDN as the future Internet architecture (FIA)  [12], [13], [14], [15], [16], [17], [18]. They have discussed about design principles of NDN compared to other FIAs, such as FIND  [19], NEBULA  [20], XIA  [21], GENI  [22], 4WARD  [23], FIRE  [24], AKARI  [25], JGN2Plus  [26], etc.

NDN, although novel and promising, has many open research challenges, like scalability problem in routing, want of support of wire-rate forwarding, and absence of a proper naming guideline for application development and network data delivery. A lot of work has been done on NDN within a very brief time span, but no survey exists till date to study about NDN architecture and its features in a contiguous and coherent manner. In this paper, we shall analyze NDN characteristic features and summarize several proposed techniques on various functionalities of NDN. We shall further compare/contrast features and functionalities of NDN and IP in a purely objective manner for the benefit of future researchers. In summary, in this paper, we make the following key contributions.

• We propose a feature tree-based taxonomy that organizes NDN key features and their relationships into a framework to help understand and classify the existing work.

• We provide a brief overview of NDN and its characteristic features and compare them with core functionalities of the Internet.

• We provide a detailed review of existing works on different aspects of NDN as per our proposed taxonomy, covering architecture and organizational structure, routing, data dissemination and retrieval strategies, and applications.

• We also discuss several open problems and identify directions for future research.

The remainder of this paper is organized as follows. In Section  2, we present the tree taxonomy of NDN in terms of the architecture, system services, and applications. In Section  3, we briefly explain the core features and functions of NDN. In Sections  4 NDN system services, 5 Applications, we provide a detailed survey and analysis of several NDN characteristic features based on our proposed taxonomy. In Section  6, we provide a functional comparison of NDN with the Internet/IP. Finally, in Section  7, we conclude the paper after discussing the challenges, open problems, and future directions of NDN research.