1089-7798 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LCOMM.2018.2864144, IEEE
Communications Letters
1
Compressed Sensing-based Cached Content Locating
for ICN
Qingyi Zhang, Xingwei Wang, Min Huang, and Jianhui Lv
Abstract—Locating the cached content accurately and timely
is critical for utilizing in-network storage in information centric
networks (ICN). However, an enormous number of cached contents
and their volatility nature make it a challenge to collect and
maintain a consistent view of the network cache information. In this
letter, we propose a novel cached content locating mechanism for
ICN, which jointly uses the bloom filter and compressed sensing to
efficiently represent the cached contents and leverages the ability of
centralized control of software-defined networking (SDN) to make
the cache information consistent in the SDN controller. Simulation
results show a large saving in announcement cost with a tolerable
decrease in locating accuracy.
Index Terms—ICN, SDN, compressed sensing, content locating,
bloom filter.
I. Introduction
I
NFORMATION centric networks (ICN) replaces IP packet
with named content and realizes name-based routing. This
change makes a fundamental benefit for the utilization of in-
network cache. To make full use of it, content requests should be
ideally forwarded to the closest nodes that cache the correspond-
ing content. Therefore, locating the cached content accurately in
time becomes a prime challenge for ICN.
Recently, [1] speeds up content retrieval via an interest
similarity relation discovery scheme based on maximal tree.
[2] combines the cache information of neighbors into a single
bloom filter (BF) and exchanges the information among 1-hop
neighbors. [3] applies deep belief network to predicting the most
popular contents and advertises them using the BF. [4] exploits
in-network caches by hash routing, in which the location of
a content is determined by a hash function. [5] proposes a
resolution-based design which employs a tracker server as name
resolution system to locate the nearest contents.
However, the previous work is faced with the following open
issues. 1) Due to the enormous number of cached contents,
collecting and maintaining the cache information of every ICN
node will cause huge overhead. 2) The cached contents are
This work is supported by the Major International (Regional) Joint Research
Project of NSFC under Grant No. 71620107003, the National Science Founda-
tion for Distinguished Young Scholars of China under Grant No. 71325002, the
National Natural Science Foundation of China under Grant No. 61572123, the
Program for Liaoning Innovative Research Term in University under Grant No.
LT2016007, and the MoE and China Mobile Joint Research Fund under No.
MCM20160201. (Corresponding author:Xingwei Wang and Min Huang).
Q. Zhang and X. Wang are with the College of Computer Science
and Engineering, Northeastern University, Shenyang 110169, China (e-mail:
wangxw@mail.neu.edu.cn).
M. Huang is with the College of Information Science and
Engineering, Northeastern University, Shenyang 110819, China (e-mail:
mhuang@mail.neu.edu.cn).
J. Lv is with the Central Research Institute, Network Technology Laboratory,
Huawei Technologies Co., Ltd., Shenzhen 518129, China.
Forwarding Engine
1
1 0 1
1
0
Communication Interface
Updater
LCIS
Sampling
Local
Cache
Communication Interface
CIB
Updater
Reconstructor
Recovery
Swap
Area
Reserved
Area
CIB
ICN
Node
ICN
Node
ICN
Node
ICN
Node
SDN
Controller
Routing Table
Cache Information
Management
Sensing
Matrix
Mapping
Operator
querying
App
1
App
2
App
n
Data Flow
updating
mapping
compressing
recovering
updating
Locating
setting
next hop
a
b1
b2
b3
c1
c2
d1
d2
Fig. 1: System overview
volatile, because contents are dynamically cached and deleted
in in-network storage. 3) The cache information should be peri-
odically exchanged among distributed ICN nodes to keep such
information consistent. However, due to the volatile nature of
cached contents, a more frequent cache information exchange is
required. What’s worse, the volume of cached contents is huge.
The information exchange will cause an explosive growth of
control traffic and slow down the speed of achieving consistency.
Although the BF can efficiently represent the cache informa-
tion, the space of BF is not fully utilized for controlling the false
positive. Therefore, we use the theory of compressed sensing
(CS) [6] to further compress the BF. CS shifts the computational
cost from encoder to decoder. We take advantage of this feature
by leveraging software-defined networking (SDN), i.e., all ICN
nodes periodically announce the encoded cache information to a
centralized SDN controller which is responsible to decode and
maintain such information. Furthermore, to make the announce-
ment meaningful, a cache-reserving mechanism is proposed
to reserve the announced cached contents, which solves the
volatility problem. Since the announced cached contents are
reserved and this information is maintained by the controller,
the consistency of the cache information is ensured.
The main contributions of this letter include: 1) We propose
an SDN-aided CS-based cached content locating mechanism for
ICN (C
3
L) that comprehensively considers the above-mentioned
three problems. 2) We jointly apply the BF and CS theory to
reduce the overhead of announcing the cache information. 3) A
new recovery method based on a pre-process for CS is proposed.
II. System Overview
A. Architecture
The structure of the proposed SDN-based ICN cached content
locating mechanism is depicted in Fig. 1, which consists of two
planes, i.e., the data plane and the control plane.