Part 1: Efficient Multimedia Delivery in Content-Centric Mobile Networks

CC Lab.
Efficient Multimedia Delivery in
Content-Centric Mobile Networks
Dept. of Information and Communications Engineering
Hankuk University of Foreign Studies, Korea
Presented By
Md Mahfuzur Rahman Bosunia

CC Lab.
Hankuk University of Foreign Studies
Outline
Background and Motivation
State-of-the-Art Work
An Architecture for CCN-based Mobile Networks
Routing and Mobility Management
Smart Base Station-Assisted Content Delivery
Conclusion
26/22/2017

CC Lab.
Background and Motivation
36/22/2017

CC Lab.
Today’s Communication
 IP address: naming an attachment point (host)
 DNS-based communication
 Point-to-point delivery
 Content Delivery Network (CDN) and Peer-to-Peer (P2P)
46/22/2017

CC Lab.
Data Traffic Prediction
5
Source: Cisco VNI February 7, 2017
• Global Mobile Data Traffic:
• Half a billion (429 million) mobile devices and connections were added in 2016.
• Global mobile data traffic grew 63% (4.4  7.2 exabytes) in 2016.
• 4G traffic accounted for 69% of mobile traffic in 2016.
• Mobile will represent 20 percent of total IP traffic by 2021.
• The number of mobile-connected devices per capita will reach 1.5 by 2021.
• Overall mobile data traffic is expected to grow to 49 exabytes per month by 2021.
• The total number of smartphones will be over 50 percent of global devices and
connections by 2021.
•
6/22/2017

CC Lab.
Challenges in Today’s Communications
66/22/2017
 Host-Centric Communications.
 Overlaying Model: Many protocols, many overlay.
 Internet Protocol: Communicate by addressing device.
 Addressing ~ Inefficient : DNS, IPv4.
 Scalability Problem: Group-based communication.
 Mobility: Achievable but not inherent.
 Security ~ Connection : IPSec, VPN, SSL.

CC Lab.
Content-Centric Networking
76/22/2017
CCN represents the third revolution in telecommunication networks….
Van Jacobson
Content Centric Networking (CCN) Network of Information (NetInf)
Data Oriented Networking
Architecture (DONA)
Publish Subscribe Internet Routing
Paradigm (Pubs/Sub)
 Contents/Data are at the center of networking.
 Name Contents not Device/Location.
 Pull-based consumer driven.
 Request DATA in the Network.
 Any node with DATA, Responds.
 Faster delivery, less delay, content-level security.
 Representative Architectures:

CC Lab.
86/22/2017

CC Lab.
File type: pdf,
Title: ccn_archi,
Author: xxx,
Organization: hufs
9
Content-Centric Architecture and Routing
 Combined Broadcast and Content Based [1]
 Publish/Subscribe architecture
 Uses two different routing protocols:
 Broadcast protocol: constructed the network topology
 Content-based Routing: Used to deliver content
Strengths of CBCB Weakness of CBCB
 Content oriented communication
 Content is delivered based on the
subscription of a content
 Overlay application
 Uses the traditional IP-based
protocols
6/22/2017

CC Lab.
State-of-the-Art Work (cont.)
10
 DONA [2]  PSIRP[3]
 Clean redesign of Internet
architecture
 Uses hierarchical name resolution.
 Uses flat and unique naming.
 Overlay of RHs
 Depends on RPs
 RPs holds the content publication and
subscription
 Map content identifier to RP
identifier
Strengths of DONA and PSIRP Weakness of DONA and PSIRP
 Resolve dependency on DNS
 Content caching
 Domain level switching
 Routing Complexity
 Not clearly defined the real
prototype
6/22/2017

CC Lab.
11
 NetInf [4]
 Uses MDHT-based name resolution
 MDHTs uses globally unique identifier.
 NR provides information regarding the content location and content
attributes.
Strengths of NetInf Weakness of NetInf
 Resolve dependency on DNS.
 Provides addition content layer to
preserve device anonymity.
 Creates hierarchical spanning tree.
 Decouples name resolution from
content routing.
6/22/2017

CC Lab.
12
 CCN [5]
 Full-fledged content centric network architecture.
 Focuses on the what not the where.
 Uses only content name to locate or retrieve content.
Strengths of CCN Weakness of CCN
 Eliminates the dependency on NRS
or DNS.
 Provides in-network data delivery.
 Internet can not directly be
integrated with CCN.
 Routing complexity and control
overhead.
6/22/2017

CC Lab.
13
Mobility Management
 JUNO [6]
 Achieved consumer and provider mobility using
a middle layer.
 Uses DHT-based Content discovery:
 Identify content rather than routing.
 Uses many overlays of applications.
Strengths of JUNO
Weakness of JUNO
 Intelligently re-configure content
delivery.
 Re-selects content source after re-
location.
 Middle-layer increase the
complexity.
 Focuses on backward compatibility.
 CCN[5], DONA [2], NetInf [4], Mobility First[5]
 Achieved mobility by simply updating content prefix and re-binding.
6/22/2017

CC Lab.
14
 Proxy-based [7], PMC [8]and Clustered CCN [9]
 Hierarchical mobility management solution.
 Root-node handles all the responsibility like interest tracking,
device tracking, and flow identification.
Strengths Weakness
 Support consumer mobility or
provider mobility.
 Reduces packets flooding.
 Control overhead.
 Deployment complexity.
6/22/2017

CC Lab.
15
Multi-Hop D2D Communication and Data Distribution
 MANET-aware CCN [12], OSPFN [13], NLSR[14], ACCC [15]
 Borrowing the concept of CCN in MANET.
 Uses link state information to make routing
information:
 Prefix advertisements.
 On-demand transmission.
Strengths
Weakness
 Simple mechanism.
 Provides pervasive content
delivery.
 Borrow the idea of MAC protocols.
 Network wide packet flooding.
 Multi-hop D2D communication [10]
 Proposed in LTE network where a distributed clustering approach is used [11].
Consumer
Provider
Interest
6/22/2017

CC Lab.
16
Multi-Source Multi-Path Content retrieval
 Proposed to show the feasibility,
applicability, and effectiveness of the
CCN.
 Defined different forwarding strategies
such as,
 Interest forwarding to multiple
content source.
 Chunk-level flow differentiation.
 CCN [5][16]  Others [17][18] [19]
 Split Content into multiple source.
 Replicate Interest into multiple
source.
 Propose a RTT-based congestion
avoidance to shape Interest rate.
 Uses ant-colony based greedy
forwarding
Strengths Weakness
 Control the Interest rate.
 Reduces packets flooding.
 Cannot utilizes the benefit like p2p.
 Does not consider bandwidth
utilization or resource optimization.
6/22/2017

CC Lab.
17
Future Edge Network
 Software Defined Networking [20]
 Provides faster data access.
 Meets the need of heterogeneity.
 Separates the control layer and data
layer.
 Network intelligence is distributed in the
programmable controller.
Strengths of SDN Weakness of SDN
 Provides greater openness and
flexibility.
 SDN is still limited to the
wired environment.
6/22/2017

CC Lab.
18
 CRAN [21], Open Flow [22], MEC [23]
 Efficient use of computation
resources.
 Controller based data delivery.
 Network intelligence is
distributed.
Strengths Weakness
 Deployments of high computational
server at the edge.
 Provides real time access at the
application level and RAN level.
 Based on the IP
architecture.
 Till now, CRAN or MEC is
not available in the wireless
base stations.
6/22/2017

CC Lab.
Challenges Present in CCN-based Mobile
Networks
 No state-of-the-art routing architecture and routing protocols to
support billions of devices in the wireless environment.
 CCN is not still well explored to provide ubiquitous solution and
seamless mobility.
 Content distribution and packet flooding is also an another open
issue.
 Reliable congestion Control.
 Provide better connectivity and faster content access.
 Customize user centric and device centric experiences.
196/22/2017

CC Lab.
Motivation of the Thesis
 Design of well-structured network architecture.
 Routing and Content Retrieval
 Efficient mapping between named content and underlying topology.
 Mobility Management
 Consumer Mobility
 Provider Mobility
 Traffic Engineering
 Reliable, congestion and flow controlled transport.
 D2D-based interactive and adaptive content Delivery
 Multimedia sharing and retrieval
 P2P manner.
 Multi-source and Multi-Path Content Delivery.
 Smart Access to the Network
206/22/2017
Original
Content “XY1”
Owner
“M”

CC Lab.
An Architecture for Content-Centric Mobile Networks
216/22/2017

CC Lab.
An Architecture for CCN-based Mobile
Networks
• Node in the Network has three functional elements:
• Content Store (CS)
• Forwarding Information Base (FIB)
• Pending Interest Table (PIT)
22
• Follows a hierarchical structure
• Valid and meaningful reasoning of each
component
 Node Model and Naming
A naming Structure
6/22/2017

CC Lab.
Networks (cont.)
23
• The proposed CCN architecture is
comprised of:
• Smart Mobile Device (SMD)
• Smart Base Station (SBS)
• Smart Router (SR)
• Content Server
• Router with CCN functionalities
and storage.
• A conjoint devices that can store, match, and
forward content and routing information. An Architecture for CCN-based
Mobile Networks
 SR
6/22/2017

CC Lab.
Network (cont.)
• Devices are able to communicate with
other directly.
• Integrated with various components to
control content delivery, QoS, and
mobility.
24
• Evolves functionalities, such as the segmentation of the content into
uniquely identified chunks, in-network content storage, and name-based
routing with backhaul network.
• SBS also contains some smart technologies, mainly including semantic
data processing technology.
Smart Mobile Device
 SMD
 SBS
6/22/2017

CC Lab.
Basic Routing
• Two types of messages are used for communication:
• Interest Packet
• Data Packet
25
The Interest packet can also hold req flag, to request the meta-
information (e.g., content size, content source description,
bandwidth, data rate).
The Data packet holds datareq flag, to indicate meta-information.
• Each node is self-dependent to assign valid, meaningful, and straightforward content
name.
• Register name in the CS with its LV value and spread the name prefix.
• Content publisher spread the name, prefix or host name to the nearby nodes.
• The hierarchical name prefix will eventually reach the SR.
• A node may attain the FIB by tracking the Data packet on the fly.
 Content Publication
 FIB Entry Manipulation
6/22/2017

CC Lab.
Basic Routing (cont.)
• Send Interest with Content
name.
• When any node receive Interest,
reinforces a lookup inside CS.
• If requested data is found then
reply back the Data, otherwise
forward the Interest packet.
26
A CCN-based Routing
 Content Request
6/22/2017

CC Lab.
Basic Routing (cont.)
• When an Interest is received,
Data packet is sent back using
the backward route.
• Remove the entry in PIT.
27
A CCN-based Routing
 Content Retrieval
6/22/2017

CC Lab.
Reliable Content Retrieval
• Uses a mathematical model, that is capable to priorities the receiver’s demands
in terms of performance criteria while choosing a face or source, e.g., available
bandwidth, transmission delay, and transmission cost.
• Let x is an instance present different estimation for a single function,
28
(3.3)
• It is defined a multi-criteria choice function AF that integrates the different
demand metrics or single criteria function of content consumer or any
intermediate device to select a best transmission path.
(3.4)
where wi is a user defined weight, reflects the importance of each
choice criteria and learned over time.
6/22/2017

CC Lab.
• Consumer send Interest with req flag.
• Provider reply meta-information with dataReq flag.
• Consumer chooses a best face.
• Provide QoS.
29
Reliable Content Retrieval (cont.)
 A Route for Content Request
6/22/2017

CC Lab.
Mobility Support
• It uses the mobility prediction of the end devices as well
as adopt the make before break approach for handover
operation.
30
A Mobility Scenario
Handover Procedure
6/22/2017

CC Lab.
Performance Evaluation
316/22/2017
• The proposed architecture was implemented using:
• NS-3, DCE [124][125]
• CCNx [5]
• Simulation includes various access networks:
• Wi-Fi
• LTE
• Wired LAN
 Simulation Environment
• Three routing metric was used:
• Hop distance: 0.6, bandwidth: 0.25,
number of flows: 0.15.

CC Lab.
Performance Evaluation (cont.)
32
Heterogeneous Access-networks
6/22/2017
• 10 MB video file was published.
• The content transfer time is much lower after the first fetching because of
the smart data caching at intermediate node uses by the CCN.
• The basic routing approach transmits Interest blindly, and therefore
sometimes it selects an overloaded route so it demands larger time to fetch
the content.
 Simulation Scenarios and Results
Content Transfer Time
1Gbps

CC Lab.
33
LTE Access Network
6/22/2017
• UE changes its position at a speed of 60 km/h. The distance between two
SBS was considered 400 m.
• When the number of UEs are increased, the proposed architecture requires
least content retrieval time than the traditional content centric architecture
because of its balanced data transmission and also the minimization of the
routing overhead.
 Simulation Scenarios and Results
Content Transfer Time

CC Lab.
Hierarchical Routing for Mobile Networks
346/22/2017

CC Lab.
Main Idea
• Goal: path creation/computation/management between data provider and
consumer.
• Assumptions
• Providers & consumers don’t know each other’s location.
• Hierarchical Routing (H-routing):
• Creates/Manages forwarding paths
• Computes the path from the publishers to the subscribers
• Operational Scope:
• Content registration
• Content search
• Content retrieval from a possible source
356/22/2017
Directed
FIB
On-demand FIB

CC Lab.
Distributed Hash Table
• A hash table allows you to insert, lookup, update and delete with object keys.
• A distributed hash table allows you to do the same in a distributed setting.
• Performance Concerns
• Load balancing
• Fault-tolerance
• Efficiency of lookups, and inserts
• Locality
• Napster, Gnutella are all the example of using DHTs.
• Look up latency, or messages for a lookup:
• O(N)
• The goal is to provide the memory, lookup latency and messages for a lookup:
•  log2(N)
366/22/2017

CC Lab.
H-Routing for Mobile CCN
• Intelligence choice of neighbors to reduce latency and message cost of routing.
• Uses Consistent Hashing SHA-1 on each node,
• An m bit identifier, here the SHA-1 is applied to the node name, called peer id (between 0 and
2m - 1).
• Can then map peers to one of 2m logical points on a circle.
• Data Name = Node Name + Hierarchical Prefix: Information Attributes
• m + n bit identifier, where n is the unique data name.
• Information attributes hole any other identifier or IP address.
• Every node knows the successor of it.
376/22/2017
N16
N32
N45
N80
N96
N112 Let m=7
N16
N32
N4
5
N80
N96
N112i ft[i]
0 96
1 96
2 96
3 96
4 96
5 112
6 16
ith entry at peer with id n is the first peer with id n+2i(mod 2m)
80+20
80+21
80+24
Network ring

CC Lab.
H-Routing for Mobile CCN (cont.)
• H-Routing, uses three-tier network hierarchy
where the SBS are in bottom layer, SR in the
middle layer, and the data storage (DC) are in
the top layer.
• At each layer forms a logical ring using the Node
name.
• Direct mapping between Node name and
Content Name.
• Before forwarding an Interest a hierarchical
lookup is performed and Interest is forwarded to
the possible Content Source.
• Different types of messages are used,
386/22/2017
Join Interest Hello Interest
Leave Interest Key Interest
Look up Interest Register Interest
The H-Routing Architecture
The Extended Node Model

CC Lab.
Hankuk University of Foreign Studies 39
• Lookup overhead: Bottom Layer
6/22/2017
Let the number of SBS is NSBS and number of UE is
NUE, in the bottom layer ring. Let MesSBS,i is the
number of sent and received messages by a SBS.
The total number of generated messages for each UE
is TotUE,j. The total number of message generated by
all SBSs and UEs are expressed as:
• Lookup overhead: Top Layer
Let NDS be the total number of sent and
received messages by a single DS. Then the
total number of generated messages TotDS,i,
for each DC, that triggers the data lookup
process is obtained as follows:
 Mathematical Correctness

CC Lab.
• Lookup overhead: Mid Layer
6/22/2017
Let the total number of SRs be NSR. Then, we can obtain the total number of sent or received
messages by all the SRs as follows:
• Lookup overhead: Total
Lookup Overhead
3 layer ring reduces the lookup packet overhead significantly.
 Mathematical Correctness

CC Lab.
41
• CCNx with NS3-DCE
• Use Cases: Three different considered
Simulation Topology
6/22/2017
Test Case 1: Data retrieved from DS
Test Case 2: Data retrieved from UE
Test Case 3: UE roles as Consumer and Producer both

CC Lab.
42
• Performance Indicator
• Average Throughput
• Interest Packet Overhead
• Interest Packet Error Rate
• All the provider publish 1MB
file after 1 min interval.
• Content consumer makes
request randomly.
Simulation Topology
6/22/2017
• In the third use case, 50% of UEs works as provider and the
remaining works as consumer.
• Evaluated the performance of Vanilla Interest Flooding-based
CCN [26] and the proposed H-Routing mechanism.

CC Lab.
43
Test Case 1
6/22/2017
In the DS-UE and UE-DS scenario, contents as transferred on the shared high speed point-to-point
line, and limited routing required.
Test Case 2 Test Case 3
In the UE-UE scenario, UE can retrieve content from another UE independently, SBS
communicates and routes with each other, and increase the Interest packet forwarding thorough
different paths. Limited shared link.
The H-Routing shows a significant improvement, because it locates the content source before the
original transmission appeared. Reduces network wide flooding, unnecessary packet transmission.
 Average Throughput

CC Lab.
Test Case 1
6/22/2017
It can be visualized that the Interest packet overhead is much higher in case of the Vanilla Interest
Flooding in CCN.
Test Case 2 Test Case 3
H-Routing reduces the Interest packet flooding because of its content source selection capability
and unicast Interest flooding toward the actual content provider.
 Interest Overhead

CC Lab.
DS-UE
6/22/2017
The Re-Interest rate of H-Routing is much shorter than the rate of CCN because of the reduction of
Interest packet flooding and packet loss due to congestion.
UE-DS UE-UE
CCN uses the different forwarding route and results in a huge traffic congestion and packet loss
and demand new Interest forwarding.
H-Routing shows the almost 1-1 Interest-Data packet mapping relation because of its proper choice
of content selection, source selection and route selection for Interest packet forwarding.
 Interest Packet Error Rate

CC Lab.
Mobility Management
466/22/2017

CC Lab.
Mobility Prediction
• UE reports the received RSS value for all the neighbor eNodeB to the serving
eNodeB.
• The eNodeB that takes the handover decision uses a utility function to make four
sets of groups of eNodeB.
476/22/2017
Let P, Q, R, and S are four sets of eNodeBs where P contains the eNodeBs those have the highest RSS value
observed, and S contain the eNodeBs with lowest RSS. Let r be the RSS observed by an UE for a candidate
eNodeB and α be the stepness. rmin ≤ rm ≤ rmax where rm is the average of all the RSS value observed from all
the candidate eNodeB, rmax is maximum RSS among the candidate eNodeBs, rmin is the minimum RSS
observed among the candidate eNodeBs.
• Then the candidate group are constructed as follows:
u(r) = 0 ∀ r ≤ rmin, u(r) = 1 ∀ r ≥ rmax and u(rm) = 0.5.
 The Received Signal Strength (RSS) Estimation

CC Lab.
• Assumptions
• eNodeB is aware of the 2-hop neighbor eNodeBs and each UE knows its own
position.
• The moving angle  and distance d:
• Let the serving eNodeB position is (Xe, Ye), the position of the UE is (Xu, Yu), and the
candidate eNodeB position is (Xn, Yn), then  can be estimated,
486/22/2017
It is considered the 120 angle is the acceptable angle towards a SBS.
So its considered as a offset of µ= 60 to normalize the  value.
• The movement Prediction Pm:
•  and d are used to estimate the movement prediction Pm as follows,
Mobility Prediction (cont.)
 Movement Prediction

CC Lab.
• To take the accurate context based decision for handover, it considered the load of
the candidate eNodeB.
• Data rate in per unit time is an indication of the load among the eNodeBs.
• eNodeB is aware of the 2-hop neighbor eNodeBs.
• Each eNodeB transmits its current data rate per sec to its two hop neighbor eNodeB.
• The serving eNodeB uses a utility function to make different group of the
candidate eNodeB:
496/22/2017
Let T, U, V, and W are four sets of eNodeBs where T contains the eNodeBs those have the larger data rate, U contains the SBSs
those have mid-level data rate, V contain the low level SBS and W contain the eNodeBs with the lowest data rate. Let x be the
data rate known by a serving eNodeB, and α be he stepness. xmin ≤ xm ≤ xmax where xm is the average of the data rate, xmax is the
maximum observed data rate, and xmin is the minimum data rate observed among the eNodeBs
• Then the group are constructed as follows:
u(x) = 0 ∀ x ≤ xmin, u(x) = 1 ∀ x ≥ xmax and u(xm)
= 0.5.
 Load Estimation

CC Lab.
• The Best Candidate Selection:
• UE reports the measurement of RSS of all the candidate eNodeBs and its own
position information to the associated eNodeB.
• Each eNodeB sends their load estimation to 2-hop neighbor eNodeB.
• Then the associated eNodeB uses the Equation 5.1 to make the group of eNodeBs
based on the RSS estimation.
• Then serving eNodeB uses the position information of the eNodeB, those contains
in the RSS group P and Q and evaluate the direction prediction using Equation 5.8.
• The serving eNodeB then considers the load estimation and makes the set of
eNodeBs group T, U, V, and W using Equation 5.9.
• The serving eNodeB considers the candidate eNodeBs inside T and compare
their direction prediction values and selects the eNodeB that is much closer to
the UEs movement direction.
506/22/2017
 Best Candidate Selection

CC Lab.
Seamless Content Retrieval
516/22/2017
 Receivers Preference Modeling or UE Mobility Prediction
• A new connection is established with the new eNodeB before breaking the
current connection.
• eNodeB follows the procedure mention in Mobility Prediction to decide
whether the UE will move to a new eNodeB or not.
• If the serving eNodeB decides the
necessity for a new eNodeB to
continue the seamless data retrieval
of the UE, it forwards the Interest
and related information to the new
eNodeB, then the new eNodeB
forwards the Interest to the most
appropriate content provider to
retrieve the content.
Seamless Data Retrieval with Consumer Mobility

CC Lab.
Seamless Content Delivery
• Let a content be denoted by d which consists of naming attributes,
e.g., location, type denoted by vd.
• Thus the content name is represented by,
526/22/2017
 Content Similarity Function: Provider Mobility
• The following formula is used to calculate the similarity, S1,2
between content item d1 and content item d2.
• where m is the number of attributes that
present the content, e.g., video, size, length,
• wi is the weight for each attribute,
• B (i, m) is a similarity function returning,
• 1 if vi
d1 = vi
d2, and 0 otherwise.
Due to mobility, if the content name is changed and the content producer receives the old named
Interest message, then it use the content similarity function to determine the Interest will be
satisfied or not.

CC Lab.
53
• Performance Indicator
• Content Transfer Time.
• Data Transmission Success
Ratio.
Simulation Topology
6/22/2017
Content Size: 4 Mbytes. The number of eNodeB is three and the number of Ues
covered by each eNodeB varies from 1~10. Each eNodeB is placed with 400m
away from each other. The mobility speed is 0~60 km/h. Each UE published the
different number of data files, several UE provide same content.
• The proposed CCN-based mobility management approach has been
evaluated and compared with the current TCP/IP based content transfer in the
LTE network and CCN-based content transfer in the LTE network.

CC Lab.
546/22/2017
 Simulation Results
The content transfer time of the mobility
prediction based approach is shorter than others
because of its efficient soft handover based
mobility management mechanism for both
consumers and producers.
The introduction of the extra buffer reduces the
chance of long transmission delay, queuing delay,
propagation delay and processing delay.
The proposed content similarity approach
increases the data availability when mobility
changes the content location.
Also the buffering capability, fast path switch,
and the handover prediction reduce the packet
loss rate.
Able to detect and differentiate losses due to
congestion, link failure, and mobility.
Content Transfer Time Content Transfer Success ratio

CC Lab.
Content-Centric D2D Communication
556/22/2017

CC Lab.
Multi-hop D2D Communication
566/22/2017
• Multi-hop relying or D2D communication has been considered as a key
technology for future wireless network.
• Enhance efficient content distribution, in order to enable mobile users to
ubiquitous access to the nearby contents.
• It also propose an efficient Interest forwarding based on the face lifetime
prediction of the devices.
• Assumptions:
• Each device is aware of its own position and moves randomly.
 No infrastructure
 Limited energy
 Mobility
 Control overhead
Unstable route
Low energy
Unreliable data
delivery
Broadcast overhead

CC Lab.
Forwarding Interest
• Face Lifetime Estimation:
• Each node holds an extra data structure the movement prediction table.
• HELLO message are used to sends location information, mobility speed, mobility
direction, and available contents.
576/22/2017
Let assume two devices A and B are within the transmission range R, the device A is at the position (XA, YB)
and is moving with a speed VA and angle A, and the device B is at position (XB, YB) and is moving with a
speed VB and angle B, where, 0 > A , B > 2. Then, LLTA,B, is measured as
• Then, the face lifetime can be estimated as follows:
For accomplishing a more realistic estimation of the LLT, an weighted average of the LLT (LLTavg) from n
recent observation are considered as follows,

CC Lab.
Popularity Based Content Distribution
586/22/2017
• Let the interest rate for content c is q(c). Then the
delivery rate d(c) for content c as follows:

CC Lab.
Main Principles and Processes
59
• Firstly, nodes need to find a face to forward Interest packet to retrieve a content.
• Each node uses the Equation 6.4 to estimate which face has the highest face
time than others, then forward the Interest packet.
• Periodically, each device estimates the Interest rate, Data rate and Content
popularity value of all contents available in a device.
• The corresponding device discards the data that has the lowest estimated
popularity value.
6/22/2017
D2D Data Transmission Scenario

CC Lab.
60
• The proposed mechanism compared
with FIFO-based traditional CCN
approach.
• Performance Metrics:
 Interest Satisfaction Rate:
A Challenged Network Topology for
Simulation
6/22/2017
In the first evaluation, it was considered
the different storage capacity of each
devices and the capacity are varied from
5 to 30.
In the second evaluation, the total
number of fetching content at each
device are increased from 5 to 30
number of different content in a 5 to 10
seconds interval.

CC Lab.
61
Interest Satisfaction Rate for Different Buffer Limit
6/22/2017
Each of the devices publishes a different content insides
it CS and make request for 5 different contents in a 5 to
10 seconds interval.
Interest Satisfaction Rate for Different Number of Interests
The proposed Interest forwarding strategy and data
distribution mechanism achieves a 75% performance
improvement in the top most popular content retrieval.
The Interest packet satisfaction rate is dropped
off as the Interest packet load is increased in
both approach.
The basic CCN used the network wide packet flooding
and conventional data buffering technique e.g., First-In
First-Out (FIFO)-based Least Recently Used (LRU) data
distribution policy.
The basic CCN shows the performance
degradation due to its unbalanced request
dissemination and content distribution called
disequilibrium techniques.

CC Lab.
A Multi-Source and Multi-Path Transport Mechanism
626/22/2017

CC Lab.
Main Idea
63
• Content retrieval from multiple sources simultaneously.
• Multi-source and multi-path efficiently utilizes network resources and reduces
communication costs, data storage and computation.
• Introduce an on-demand and dynamic multi-source interest forwarding strategy.
• CCNx prototype is enhanced. Develop two applications named
cclabreceiver and cclabprovider for content retrieval and content
delivery from multiple sources.
 Content retrieval from different possible sources
6/22/2017

CC Lab.
Problem Modeling and Motivation
64
• Let there are n end consumer devices and m content provider, each of them is
independent. Same content are available in m providers.
• The Interest rate i and data delivery rate µi.
• This scenario can be mapped as an M/M/m queuing model.
6/22/2017
D2D Data Transmission Scenario
Let a consumer is sending Interest packet at a rate of 15 chunk/sec, and each
chunk takes 0.05 sec to retrieve from a provider. The mean number of Interest in
the system Ls, mean waiting time to satisfy Interest in the system Ws, mean
waiting time in the buffer Wq, mean waiting number in the buffer Lq, Service rate
µ = 1/0.05 = 20 and utilization ρ =  / µ = 15 /20 = 0.75
Now consider the same content is available in two provider, then

CC Lab.
Combine Metric Design
65
• We assume a scenario where same content is available from multiple content
source. The eNodeB selects the best UE content provider.
• Content availability metric, a boolean function of B(u∈N,c) that returns 1 if
UE, u has the content c, and 0 otherwise
• Response Time Metric:
• Interest Load Metric:
• Reliability Metric:
• Aggregate Metric:
 Multi-source Interest Forwarding
6/22/2017
where Xn,c denote the measurement of the number
of pending Interest for chunk of content c and Xn,w
denote the measurement of the number of pending
Interest for all other content towards n indexed
UE in the time of t seconds.

CC Lab.
Multi-Source Interest Forwarding
66
 Operation in Details
6/22/2017
Content is published inside the provider UE
and registered inside the eNodeB. eNodeB
keep track of all contents respect to each UE.
eNodeB has an additional functional elements
called Response Time Window(RTW). RTW
contains the request time and response time
Decision Process for Multi-Source Interest Forwarding
Operational Details of Multi-source Multi-path Transport Mechanism

CC Lab.
67
• Modified CCNx with NS3-DCE
• The proposed mechanism compared
with traditional CCN approach.
 Content Retrieval Time
 Throughput
 Interest-Data Ratio
Network Topology for Simulation
6/22/2017
Assigned each weight with the equal value of 1/4.
Each content provider publishes 5 different content file
in the whole simulation time, and same content is
published from different provider at the same time.
For each simulation the number of content provider are
increased.

CC Lab.
686/22/2017
Interest-Data rate
Due to the best performer forwarding strategy of conventional CCN, each Interest chunk always goes to the best
content provider in the same connection route and result in a huge delay and congestions during data transmission.
Average throughput observed inside a consumerContent Retrieval Time
Multiple parallel slots for Interest forwarding, and multisource Interest forwarding of the eNodeB, less retrieval time
is required, and shows the stable behavior in data rate achieved over time.
When the number of providers, as well as the number of published content increases in the network, the Interest
generating rate also increase. Our proposed approach shows a reasonable increase in the Interest packet forwarding
rate.

CC Lab.
Smart Base Station-Assisted Content Delivery
696/22/2017

CC Lab.
Future Cellular Networks
70
 Objective
6/22/2017
3GPP Core or Conventional LTE
Beyond 4G with CCN

CC Lab.
Smart Base Station (SBS)
71
 Protocol Stack and Communication Architecture
6/22/2017
Forwarding module inside SBS
Protocol Stack for UE and SBS
• SBS with IP and CCN stack.
• SBS can process CCN Interest and Data Packet independently.
Applications over Extended Protocol Stack
Communication Architecture among the SBSs

CC Lab.
SBS Functional Architecture
726/22/2017
 eNB has the IP stack
 eNB has the routing capability
 eNB can work independently
without PGW
 Direct IP packet communication
between UE and eNB
Direct mapping between Node
name and Content Name
Computation: Mobility, Profiling,
Content Sharing, Learning are
performed in Cloud
Distributed Architecture among the SBSs
Functional Architecture of SBS

CC Lab.
Content Distribution
736/22/2017
 Learning and Preference Based
• Learning Module will collect the UE data per SBS and profile the content in
which SBS it is mostly suitable to store.
• Uses the very popular and basic Collaborative filtering technique.
Where rSBSx,p is the total no of UE or total number of frequency in SBS SBSx, downloaded the content
p, and the rSBSx is the average frequency for SBS SBSx of all the content it retrieved.
Where ax (i,j): is the total number of common attribute between content i and content j, wx is
the assigned weight for different attribute, and bx (i,j) is the maximum number of common
characteristics between content i and content j.
Where m is the top m SBS those
have higher similarity SBS s, in
the Table 8.2, and m is the top m
contents those have higher
similarity to Content c.

CC Lab.
74
• CCNx with NS3-DCE in LTE and Modified LTE.
 Content Retrieval Time
6/22/2017
The scenario topology consists of five SBS, and each
SBS serves 10 UEs that are randomly deployed to move
freely with the random walk mobility model with a
coverage of 700 meter.

CC Lab.
Hankuk University of Foreign Studies 756/22/2017
Reduce 2ms time per 1Kbyte IP packet transmission to
eNB. Downlink time same to conventional LTE
 Case 1: Provider: eNB, Consumer: UE
1st time
content
retrieval; No
caching
Our proposed SBS achieves a 300% improvement in
content delivery from the SBS
Performance Significance
 Case 2: Content Retrieval from Cloud Server
Reduce packet mapping overhead between PGW and UE (consumer) in advanced mode.

CC Lab.
Hankuk University of Foreign Studies 766/22/2017
5*10 different contents are published in the cloud server. SBS buffer capacity is set to 5.
Contents initial popularity are assigned and each contents are requested based on the Zipf
distribution as follows:
 Case 3: Popularity Based Data Distribution
The content are requested in different consumer in a 3~5 sec
interval, so consumer can retrieve the most popular content
form the SBS directly rather than retrieving from the cloud
server.
 Use Case 3: Provider: UE, Consumer: UE
Data is transmitted directly between two serving SBS.
Two separated data flow
between UE and PGW in
conventional LTE.

CC Lab.
Conclusion
776/22/2017

CC Lab.
Conclusion
786/22/2017
 CCN is an emerging paradigm for Future Internet, which able to provide:
 more efficient, faster, scalable, secure, collaborative, location and medium independent
data transmission
 There are multiple challenges and problem domains are considered in this
dissertation and evolved with new, promising and consistent solution:
 Possible future network architecture for mobile network.
 Efficient data retrieval framework, called H-Routing.
 Novel mobility management for seamless content delivery.
 Efficient communication framework and balanced data distribution model for data delivery
in D2D environment.
Presents a content-centric multi-source and multipath path data transmission mechanism.
Define the functional structure of content-centric Samar Base Station.

CC Lab.
Conclusion (cont.)
796/22/2017
 In summary, the whole dissertation include the following benefits in data
communication:
Reduces time to content delivery and content access.
 Provide seamless and ubiquitous transmission experience by providing users or devices to
easily transmit or receive data to/from heterogeneous devices, base stations, and content
providers simultaneously.
 Avoids congestion in the network in the hop-by-hop manner and reduces data delivery
delay and data losses by using multiple path and multiple content source..
Increase network efficiency by minimizing redundant control information with using
approximate routing for content searching and interest forwarding.
Provide emerging data transmission services and a lot of applications by facilitating the fast
and easy data access in the wireless network and exploiting group based-multicast,
autonomous or ubiquitous communication, and context-aware decision.
Allows end users to express their intent in data communications, e.g., prioritize data.

CC Lab.
Conclusion (cont.)
806/22/2017
 Future Work:
H-Routing forms a logical ring, and their communication approach is distributed, and
highly dynamic. All nodes in a ring as typically indistinguishable in functionality.
Requires a lot of maintenance message. Duplicity probability in SHA-1.
 This is the future plan to adopt the there dimensional mobility management of the devices.
 More advanced mechanism to provide location aware and context aware data delivery.
Develop a hop-by-hop Interest shaping and controlling.
Provide end users to express their intent in communications, e.g., prioritize data.
Our future work is to explore the practical applications of the proposed framework in order
to verify its applicability and usefulness.

CC Lab.
Thanks!
816/22/2017
CC Lab.

CC Lab.
The List of Publications
826/22/2017
1. MR Bosunia, K Hasan, NA Nasir, S Kwon, SH Jeong, "Efficient data delivery based on content-centric
networking for Internet of Things applications," International Journal of Distributed Sensor Networks (SCIE) 12 (8),
2016.
2. M. Bosunia, A. Kim, D. Jeong, C. Park, Seong-Ho Jeong, "Enhanced Multimedia Data Delivery based on
Content-Centric Networking in Wireless Networks," Journal of Appl. Math and Info. Sci. (SCIE), 9.2L, pp. 579-589,
2015.
3. Mahfuzur Rahman Bosunia, Nazib Abdun Nasir, Kamrul Hasan, Seong-Ho Jeong, "A Multi-Source and Multi-
Path Transport Mechanism for Content-Centric Mobile Networks," International Journal of Distributed Sensor
Networks (SCIE), Under Review.
4. Mahfuzur Rahman Bosunia, Seong-Ho Jeong, "Machine-to-Machine Content Retrieval in Wireless Networks," in
preparation.
5. Mahfuzur Rahman Bosunia, Seong-Ho Jeong, "An Efficient Content Delivery Mechanism for Mobile
Communications," in preparation.
6. Mahfuzur Rahman Bosunia, Seong-Ho Jeong, "Efficient mobility support for content delivery in mobile
communications," International Conference on Information Networking (ICOIN), pp. 118-121, January 2017.
7. Seonghyuck Kwon, Mahfuzur Rahman Bosunia, Seong-Ho Jeong, "Optimal path based content delivery in
content-centric mobile networks," KICS Winter Conference, March 2017.

CC Lab.
The List of Publications (cont.)
836/22/2017
8. Mahfuzur Rahman Bosunia, Seonghyuck Kwon, Seong-Ho Jeong, "A CCN-based multi-source and multi-path
transport mechanism for wireless mobile networks," International Conference on Information Networking (ICOIN),
pp. 30-34, January 2017.
9. Mahfuzur Rahman Bosunia, Seong-Ho Jeong, "Provider Mobility Support in Content-Centric Wireless Network,"
Summer Workshop on Computer Communications (SWCC), August 2016.
10. Mahfuzur Rahman Bosunia, Seong-Ho Jeong, "Content-centric distribution in wireless networks," Ubiquitous
and Future Networks (ICUFN), July 2016.
11. Nazib Abdun Nasir, Minsub Lee, Mahfuzur Rahman Bosunia, Seong-Ho Jeong, "Performance analysis of
content-centric networks with mobility support," Information and Communication Technology Convergence (ICTC),
October 2015.
12. Minsub Lee, Nazib Abdun Nasir, Mahfuzur Rahman Bosunia, Seong-Ho Jeong, "Performance evaluation of
content-centric LTE networks," Information and Communication Technology Convergence (ICTC), October 2015.
13. Mahfuzur Rahman Bosunia, Chanhong Park, Seong-Ho Jeong, "A New Routing Protocol with High Energy
Efficiency and Reliability for Data Delivery in Mobile Ad-Hoc Networks," International Journal of Distributed
Sensor Networks, vol. 2015, no. 12, January 2015.
14. Mahfuzur Rahman Bosunia, Anbin Kim, Daniel P. Jeong, Chanhong Park, Seong-Ho Jeong, "Efficient Data
Delivery based on Content-Centric Networking," BigComp, pp. 300-304, January 2014.
15. Mahfuzur Rahman Bosunia, Anbin Kim, Chanhong Park, Seong-Ho Jeong, "An energy-aware and error-resilient
routing protocol for content delivery in mobile ad hoc networks," Information and Communication Technology
Convergence (ICTC), October 2013.

CC Lab.
References
846/22/2017
[1] A. Carzaniga, A. L. Wolf, C. A. Carzaniga, Er. L. Wolf, "A Routing Scheme for Content-Based Networking,"
IEEE INFOCOM, 2003.
[2] T. Koponen, M. Chawla, B.-G. Chun, A. Ermolinskiy, K. H. Kim, S. Shenker and I. Stoica, "A Data-Oriented
(and Beyond) Network Architecture," Proceedings of SIGCOMM, August 2007.
[3] C. Dannewitz, D. Kutscher, B. Ohlman, S. Farrell, B. Ahlgren, and H. Karl, "Network of information (netinf)-an
information-centric networking architecture," Computer Communications, 2013.
[4] V. Jacobson, D. K. Smetters, J. D. Thornton, M. F. Plass, N. H. Briggs, and R. L. Braynard, "Networking named
content," 5th ACM CoNEXT, 2009.
[5] S Nelson, G Bhanage, D Raychaudhuri, "GSTAR: Generalized storage-aware routing for MobilityFirst in the
future mobile Internet," ACM MobiArch, 2011.
[6] G. Tyson, A. Mauthe, S. Kaune, P. Grace, and T. Plagemann, "Juno: An adptive delivery-centric middleware," in
Proc. 4th Intl. Workshop on Future Media Networking (FMN), 2012.
[7] Bolla, R. Rapuzzi, and M. Repetto., "A user-centric mobility framework for multimedia interactive applications,"
In Proceedings of the 6th international conference on Symposium on Wireless Communication Systems (ISWCS'09).
2009.
[8] Dookyoon han, Munyoung Lee, Kideok Cho, Ted Taekyoung Kwon, Yanghee Choi, "Publisher mobility support
in content centric networks," In Proc. of ICOIN, pp. 214-219, 2014.
[9] Muhammad Saad, Kwangsoo Kim, Seungoh Choi, Byeong-hee Roh, "Cluster-based Mobility support in Content-
centric Networking," Research Notes in Information Science (RNIS), pp. 14: 441-444, 2013.

CC Lab.
References (cont.)
856/22/2017
[10] Mao G., Zhang Z., Anderson B. ., "Cooperative content dissemination and offloading in heterogeneous mobile
networks," IEEE Transactions on Vehicular Technology, 2015.
[11]. Del Carpio L. F., Dowhuszko A. A., Tirkkonen O., Wu G. ., "Simple clustering methods for multi-hop
cooperative device-to-device communication," Vehicular Technology Conference (VTC), May 2015.
[12] Sharma P., Souza D., Fiore E., Gottschalk J., Marquis D. ., "A case for MANET-aware content centric
networking of smartphones," 13th IEEE International Symposium on a World of Wireless, Mobile and Multimedia
Networks (WoWMoM), June 2012.
[13] L. Wang, A. Hoque, C. Yi, A. Alyyan, and B. Zhang., "OSPFN: An OSPF based routing protocol for named data
networking," University of Memphis and University of Arizona, Tech. Rep, 2012.
[14] A K M Mahmudul Hoque, Syed Obaid Amin, Adam Alyyan, Beichuan Zhang, Lixia Zhang, and Lan Wang.
2013., "NLSR: named-data link state routing protocol," ACM SIGCOMM workshop on Information-centric
networking, 2013.
[15] SE El Khawaga, AI Saleh, HA Ali., "An Administrative Cluster-based Cooperative Caching (ACCC) strategy
for Mobile Ad Hoc Networks," J Netw Comput Appl, pp. 54-76 2016.
[16] Mao G., Zhang Z., Anderson B. ., "Cooperative content dissemination and offloading in heterogeneous mobile
networks," IEEE Transactions on Vehicular Technology, 2015.
[17] D. Rossi and G. Rossini., "Caching performance of content centric networks under multi-path routing (and
more)," In Technical Report - Telecom ParisTech, 2011.
[18] A.Z. Khan, S. Baqai, and F.R. Dogar, "QoS Aware Path Selection in Content Centric Networks," Proc. of IEEE
Int’l Conf. on Communications (ICC), pp. 2645-2649, June 2012.

CC Lab.
References (cont.)
866/22/2017
[19] F. Zhang, Y. Zhang, A. Reznik, H. Liu, C. Qian, C. Xu, " Providing explicit congestion control and multi-
homing support for content-centric networking transport," Computer Communications, Volume 69, pp. 69-78,
September 2015.
[20] B. A. A. Nunes, M. Mendonca, X.-N. Nguyen, K. Obraczka, and T. Turletti, "A survey of software-defined
networking: past, present, and future of programmable networks," IEEE Communications Surveys & Tutorials, vol.
16, no. 3, pp. 1617-1634, 2014.
[21]. R Wang, H Hu, X Yang, "Potentials and Challenges of C-RAN Supporting Multi-RATs Toward 5G Mobile
Networks," IEEE Access. 2, pp. 1187-1195, 2014.
[22] A.Ooka, S. Ata, T. Koide, H. Shimonishi, and M. Murata, "OpenFlowbased content-centric networking
architecture and router implementation," Future Network and Mobile Summit, 2013.
[23] "Mobile Edge Computing Standard Portal," May 2016, https://goo.gl/0CquZr.
[24] "The NS-3 Network Simulator," http://www.nsnam.org.
[25] "Direct Code Simulaion," https://www.nsnam.org/overview/projects /direct-code-execution/.
[26] E. Baccelli, C. Mehlis, O. Hahm, T. Schmidt, and M. Wählisch, "Information centric networking in the IoT:
experiments with NDN in the wild," International conference on Information-centric networking, pp. 77 -86, 2014.

Part 1: Efficient Multimedia Delivery in Content-Centric Mobile Networks

More Related Content

What's hot (19)

Similar to Part 1: Efficient Multimedia Delivery in Content-Centric Mobile Networks (20)

Recently uploaded (20)

Part 1: Efficient Multimedia Delivery in Content-Centric Mobile Networks