没有合适的资源?快使用搜索试试~ 我知道了~
以信息为中心的网络调查:基本原理,设计和争论
资源推荐
资源详情
资源评论
China Communications • July 2015
1
self-certifying versus hierarchical names, and
edge versus pervasive caching. We hope this
survey helps clarify some mis-understandings
on ICN and achieve more consensuses.
Keywords: information-centric networking,
content-centric networking, future internet
architecture, named-data networking, pub-
lish-subscribe
I. INTRODUCTION
The Internet design is dated back to 1960s and
‘70s when resource sharing was the primary
goal. Some scarce and expensive devices like
card readers, high-speed tape drives and com-
puters were hosted on the limited number of
sites, and shared by the community. Internet
Protocol (IP)[1], which names the attachment
point, was designed to support the resource
sharing in the mainframe era. However, the In-
ternet has evolved to a very different one from
what it was. On one side, according to Cisco’s
VNI report[2], the compound annual growth
rate of network trafc is anticipated to be 29%
during 2011-2016, and traffic alone will ac-
count for 86% of all the trafc in 2016. Some
powerful cloud/service platforms are built for
the large scale data distribution. On the other
side, resources (electronic content, computing,
Abstract: The basic function of the Internet is
to delivery data (what) to serve the needs of all
applications. IP names the attachment points
(where) to facilitate ubiquitous interconnec-
tivity as the current way to deliver data. The
fundamental mismatch between data delivery
and naming attachment points leads to a lot of
challenges, e.g., mapping from data name to
IP address, handling dynamics of underlying
topology, scaling up the data distribution, and
securing communication, etc. Information-
centric networking (ICN) is proposed to shift
the focus of communication paradigm from
where to what, by making the named data the
first-class citizen in the network, The basic
consensus of ICN is to name the data inde-
pendent from its container (space dimension)
and session (time dimension), which breaks
the limitation of point-to-point IP semantic. It
scales up data distribution by utilizing avail-
able resources, and facilitates communication
to fit diverse connectivity and heterogeneous
networks. However, there are only a few con-
sensuses on the detailed design of ICN, and
quite a few different ICN architectures are
proposed. This paper reveals the rationales of
ICN from the perspective of the Internet evo-
lution, surveys different design choices, and
discusses on two debatable topics in ICN, i.e.,
A Survey on Information-Centric Networking:
Rationales, Designs and Debates
JIANG Xiaoke
1,2,3
, BI Jun
1,2,3
, NAN Guoshun
4
, LI Zhaogeng
1,2,3
1
Institute for Network Sciences and Cyberspace, Tsinghua University
2
Department of Computer Science and Technology, Tsinghua University
3
Tsinghua National Laboratory for Information Science and Technology
4
State Key Laboratory of Networking and Switching Technology, BUPT
INFORMATION-CENTRIC NETWORKING
China Communications • July 2015
2
For those applications, it is the data that
interests people, instead of where the loca-
tion is, or which device provides the data.
For example, a driver may be interested in
the distance (data) from his/her car to the
nearby cars instead of those car themselves,
a housekeeper may want to know the tem-
perature (data) in a house without caring
which sensor provides that data if there are
multiple.
A clean-slate method to address the above
issues is abandoning the IP paradigm which
was designed for resource sharing on the lim-
ited number of sites 50+ years ago, replacing
where with what to ship bits. Hence, Infor-
mation-centric Networking (ICN) is proposed
by naming the data directly, which breaks the
limitation of point-to-point IP semantic. ICN
scales up the data distribution by utilizing
redundant resources, and facilitates secure, ef-
cient and exible data delivery to t diverse
connectivity and heterogeneous networks.
Naming the data is fundamental idea of
ICN, not complete design of an architecture.
How to create a network architecture based
on named data? An ICN architecture, besides
naming the data, should support another two
functions: 1) retrieving target data, and 2) se-
curing the data.
Extensive studies have been done in ICN
research, and quite a few ICN achitectures
have been proposed so far due to lack of con-
sensuses on the design. Here we list some of
these proposals: Data-Oriented Network Ar-
chitecture (DONA)[3], Publish Subscribe In-
ternet Technology (PURSUIT)[4], and its pre-
decessor Publish-Subscribe Internet Routing
Paradigm (PSIRP)[5], Network of Information
(NetInf)[6], [7], which was initially conceived
as Architecture and Design for the Future
Internet (4WARD)[8], and evolved further
as Scalable and Adaptive Internet Solutions
(SAIL)[9], Named-Data networking (NDN)
[10], which has its roots from an earlier proj-
ect, Content-Centric Networking (CCN)[11].
More proposals and detailed design survey can
be found in the existing literatures[12], [13],
[14], [15], [16].
storage, etc) are not only spread to client sides
connected with stationary attachment point
as in PC era, but also distributed everywhere,
including different devices with diverse con-
nectivity or even in heterogeneous networks,
such as mobile phones, wearable devices,
sensors, vehicles, and satellites. IP, which is
now underpinning such an cyberspace, has in-
deed exceeded designer’s expectations. After
all, even the designers of the Internet had not
envisioned the myriad ways in which it is used
today.
However, end users essentially care about
“what” rather than “where”, and the basic
function of the Internet is to ship data to serve
the needs of applications. When IP, focusing
on talking to whom (where), is applied to de-
liver data (what) in current era, the fundamen-
tal mismatch leads to a lot of issues, including
but not limited to the followings:
•
Data is forwarded following the specific
path defined by the routing spanning tree,
without the ability to utilize extra interfac-
es, multiple end-to-end paths or redundant
data replicas. This prevents data providers
from scaling up the data distribution.
•
IP, as originally designed, provides no secu-
rity support. Solutions were added later to
secure the session, e.g., TLS, IPSec. How-
ever, these solutions only provide transient
trust – the trust is valid for the two end-
points exclusively (space constraint) during
the session period only (time constraint).
•
Naming the attachment point leads to
binding with underlying identier (layer-2
identier), which makes it hard to support
multi-homing and handle dynamic change
of underlying topology, such as mobility
support, ad-hoc scenarios. This is especially
true for some heterogeneous networks, e.g.,
sensors network, vehicle network, delay-
torrent network (DTN), wherein it is rarely
possible to build steady end-to-end channel
to ship bits.
•
IP is ill-fited for some new applications
where location or even device identity is
not important, such as Internet of Thing
(IoT), sensor network, vehicle network.
After briefly intruding
and comparing dif-
ferent design choices
adopted by major ICN
proposals, this paper
presents the most
debatable topics in
ICN, including naming
structure and caching
location.
China Communications • July 2015
3
e.g., IP multicast[17], multipath TCP[18], mul-
tipath routing[19], stream control transmission
protocol (SCTP)[20], Datagram Congestion
Control Protocol (DCCP)[21], and transport
next-generation (Tng)[22]. All of these works
try to break through the naming semantics
of IP, and use the IP namespace (together
with port and sequence number) to identify
something else, e.g., communication groups,
point-to-point paths, or message-oriented data
chunk. However, these piecemeal solutions
have not been deployed globally due to some
critical reasons, like inter-working with the
existing IP systems, lack of evident business
incentive, etc. On the other hand, overlay
systems are widely used to improve the per-
formance over IP, e.g., CDN, p2p, application-
layer multicast (ALM)[23]. However, the
overlay systems are far from perfection due to
the mismatch between the application-layer
goals and point-to-point semantic of under-
lying protocols. Overlay systems suffer from
the trust in underlying networks, handling
heterogeneity of users, providing resilience,
higher path stretch and a high link stress, etc.
For example, CDN has to fool the end users
that they are talking with the intended entity
in some hack way, while the truth is the sur-
rogates that they are connecting to, are owned
by CDN providers. The third-party role of sur-
rogates becomes fundamental conict of trust
management, and leads to potential risks for
HTTP Secure (HTTPS) based communication
[24]. P2p file-sharing system relies on appli-
cation-layer routing, which is largely indepen-
dent from the Internet routing and topology.
Thus, it leads to a lot of unwanted inter-do-
main traffic, and starves other applications,
such as web trafc[25].
A fundamental issue of patches and overlay
systems is that, each of them is a piecemeal
solution of one specific problem. The inef-
ficiency of point-to-point communication,
failure of piecemeal solutions, motivate re-
searchers to rethink the Internet architecture
in a “clean-slate” way. A basic observation
is that end users essentially concern the data
they desired without caring where the data lo-
In this survey, we give an insight into the
ICN and its design, instead of the design de-
tails or recognized challenges in different pro-
posals. The rest of this paper is structured as
follows. In Section II, we explain the necessity
of ICN network architecture, together with
efforts (and failures) of improving data deliv-
ery over IP. Section III analyzes the required
functions of ICN (naming, retrieving and se-
curing the data) as well as the possible design
choices from the perspective of network ar-
chitecture. Section IV presents debates on the
fundamental design, i.e., self-certifying versus
hierarchical names, edge versus pervasive
caching, which are hot topics in ICN research
community. Finally, we summarize the survey
in Section V.
II. MOTIVATIONS OF ICN
IP is being used for data delivery. To over-
come the issues caused by IP’s shortcomings
as mentioned in preceding section, researchers
have invented many xes.
•
Large scale data distribution is being over-
layed on top of the underlying IP network
topology. End users request data by name,
e.g, URL, instead of connecting given de-
vices first. The data is returned from the
device picked by the overlay infrastructure
instead of the one chosen by end users. Do-
main names are merely aliases of contain-
ers, but refer to data. This is especially true
in content delivery network (CDN), and
peer-to-peer (p2p) system.
•
Heterogeneous networks, such as sensor
network, vehicle network, and satellite
network, are isolated from the Internet.
Gateway is usually required to ensure the
communication between two heterogeneous
network. Furthermore, dedicated protocols
are created for heterogeneous networks.
What is more, a lot of works have been
done to improve the performance of the data
delivery over IP, and these studies can be di-
vided into patch and overlay fashion.
Some researchers developed incremental
solutions l to better support data distribution,
剩余11页未读,继续阅读
资源评论
weixin_38571992
- 粉丝: 1
- 资源: 939
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功