Future Internet Architecture: Content Centric Networks

1. CS540: NETWORK ARCHITECTURE PROF: YOUNGHEE LEE
NAMED DATA
NETWORKING
PRESENTED BY MESHINGO JACK

2. Papers
1. Jacobson, V et al (2009). Networking Named Content
2. Grassi, G et al (2014). VANET via Named Data Networking

3. CCN
INTRODUCTION
CCN NODE MODEL
TRANSPORT
ROUTING
CONTENT BASED SECURITY
PERFORMANCE EVALUATION

4. INTRODUCTION

5. Background
● Traditional TCP/IP Architecture
○ Built to solve resource sharing issues
● Use of IP address
○ IP packets contain two identifiers
■ IP address for the source
■ IP address for the destination

6. Issues
● People value the internet for WHAT content it contains
HOWEVER
● Communication is still in terms of WHERE
Source:Cisco VNI: Forecast and Methodology, 2015–2020

7. Current Challenges
● Availability
● Security
● Location-dependence

8. REPLACEMENT OF “WHERE” WITH “WHAT”

9. Introduction
● Content-Centric Networking (CCN) : networking paradigm centered
around content distribution rather than host-to-host connectivity.
● This change from host-centric to content-centric has several attractive
advantages, such as:
● network load reduction
● low dissemination latency, and
● energy efficiency.

10. Benefits of CCN
● Content caching: reduce congestion and improve delivery speed,
● Simplicity: in configuration of network devices, and
● Security : building security into the network at the data level.
Source: http://networking.khu.ac.kr/gallery/layouts/net/research/res11.htm

11. CCN Concept
TCP/IP CCN Protocol Stack

12. CCN Protocol Stack
● Strategy Layer:
○ Dynamic optimization choices required to best exploit multiple
connectivities under changing conditions
● Security Layer:
○ CCN secures the content itself
○ Avoids host based vulnerabilities

13. Similarities & Differences between CCN & IP
Similarities Differences
• Both architectures share the same hourglass
shape, with the IP/NDN layer as the narrow waist.
• Both send datagrams.
• Both follow end-to-end principle.
• Both use their own namespace for data delivery
(i.e. IP uses IP addresses to deliver datagrams
between IP nodes; NDN uses the application
name space to deliver datagrams between NDN
nodes)
• CCN secures the content while IP secures the
connections
• They use a different namespace: IP address v.s. Name.
• NDN includes a security primitive directly at the narrow
waist (every Data packet is signed).
• IP sends packets to destination addresses; NDN uses
Interest packets to fetch Data packets.
• IP (by definition) has a stateless data plane. NDN has a
stateful data plane. Together with the forwarding strategy,
this stateful data plane offers NDN networks a variety of
desired functions

14. CCN NODE MODEL

15. CCN NODE MODEL
BROADCAST INTEREST OVER
AVAILABLE CONNECTIONS RESPONSE
Packet Types:
● Interest
● Packet
● CCN communication is consumer driven

16. CCN NODE MODEL
● Broadcasting through various interfaces
● Data is transmitted only in response to an Interest and consumes that
Interest
● Data satisfies an Interest if ContentName in the Interest is a prefix of that
in the Data
● When a packet arrives on a face a longest-match lookup is made
● Allows dynamic content generation through the use of active names

17. BASIC OPERATION OF CCN
1. A packet arrives on a face [interface]
2. Longest-match look-up is performed on its name
3. An action performed based on the result of the lookup

18. CCN FORWARDING ENGINE MODEL
1. FIB:
a. forwards interest packets
towards potential sources of
matching data
b. Allows multiple sources for
data
c. Multiple output faces
2. Content Store:
a. caching functionality;
b. each packet can be used by
other consumers
3. PIT
a. Keeps track of Interests
forwarded upstream
towards content sources

19. TRANSPORT

20. TRANSPORT
● Operates on top of unreliable packet delivery services
● Loss/ damage of data in transit
○ Mobility
○ Ubiquitous computing
● Provision of reliable & resilient delivery
○ Senders are stateless
○ Consumer retransmits unsatisfied Interest
● Reliability & flow control
○ Flow balance: Retrieval of one packet per Interest
○ CCN flow balance maintained at each hop unlike TCP
○ Use of LRU memory ( cache)

21. TRANSPORT
● Sequencing
○ Uses a hierarchical naming structure
○ Names are made of various components
● Rich connectivity, mobility & strategy
○ Takes advantages of multiple interfaces on machines
○ Rapidly changing connectivity
○ Multiple connectivity through per FIB- entry face list
● Simultaneous connectivity

22. ROUTING

23. ROUTING
● Works with existing routing protocols
○ Intra domain routing [IS-IS & OSPF]
○ Inter-domain routing
● Automates routing infrastructure protection

24. SECURITY

25. CONTENT BASED SECURITY
● Protection and trust level embedded within the content rather than
connections in IP networking
● Authentication of content with digital signatures
○ content, routing, policy information
● Private content is encrypted
● Provides end to end security between content publisher and content
consumer
○ No one size fits all for trust model
● CCN security model: SDSI/SPKI
○ Model keys are mapped to identities via controlled namespaces
● Implementation of Policy Based Routing

26. EVALUATION

27. EVALUATION
BULK TRANSFER PERFORMANCE
DATA TRANSFER
EFFICIENCY
● CCN performance
comparable to TCP
● However it is lower due
to its larger header
overhead
● TCP throughput: 90%
● CCN throughput: 68%

28. EVALUATION
CONTENT DISTRIBUTION EFFICIENCY
● To measure sharing performance
○ Compare the total time taken
to simultaneously retrieve
multiple copies of a large data
file (6MB) over a network
bottleneck using TCP and
CCN.
● With a single sink, TCP's better
header efficiency allows it to
complete faster than CCN.
● But as the number of sinks
increases, TCP's completion time
increases linearly while the CCN
performance stays constant.

29. CONCLUSION
● Content is the focus as opposed to host to host connectivity
● CCN follows IP design principles but uses named content
● Simple and scalable architecture
● Enhanced security, delivery efficiency & fault tolerance
● CCN is useful for both content distribution & point to point protocols

30. V-NDN
INTRODUCTION
V-NDN: DESIGN & PRINCIPLES
DEMONSTRATIONS
V2V COMMUNICATION AT SCALE
CONCLUSION

31. INTRODUCTION

32. INTRODUCTION
● Wide range of wireless interfaces available in
modern cars
● Cars should be able to choose the best available
interface or use multiple in parallel
Power Line Communication

33. ISSUE
● Cars mostly connected to the internet via Cellular Networks only
● Two ways of connecting vehicles:
○ Vehicle to Infrastructure communication (V2I)
○ Vehicle to Vehicle Communication (V2V)
■ Usage is limited to one hop communication for collision
prevention only
● Limitation of TCP/IP in enabling the use of various applications for
V2V communication

34. PROPOSED SOLUTION
● Use of Named Data Networking (NDN) to address VANET
challenges
● Benefits of naming data:
○ Decouples communication from specific interfaces and
endpoints
○ Enable vehicles to use any available interfaces and fetch data
from any other node when there is physical connectivity
● In this paper a prototype of Vehicular NDN (V-NDN) was designed
and implemented

35. V-NDN DESIGN & PRINCIPLES

36. NDN
● NDN Data Structures:
○ Content Store (CS)
○ Pending Interest Table (PIT)
○ Forwarding Information Base (FIB)

37. V-NDN NETWORK
DATA CONSUMER
DATA PRODUCER
DATA FORWARDER
DATA MULE

38. V-NDN
● Great enabler for vehicle networking, HOWEVER,
● Modifications to NDN operations are required for VANET
environment
○ PIT: should be able to cache all received data regardless of
whether it has a matching PIT entry or whether it needs data
for itself
○ Caching strategy enables rapid dissemination of data in
highly dynamic environments
○ The data can be carried by the car even if there is no
connectivity

39. IMPLEMENTATION ● NDN Daemon:
○ core capabilities through maintaining data
structures
○ Name prefix matching
○ Packet forwarding
● NDN Local Faces
○ Support application registration, Interest request by
consumers and content delivery.
○ Use of IEEE 802.11 in ad hoc mode for V2V &
provide interface with LAL to support Wi-Fi
broadcast
● NDN Network Faces:
○ Provides adaptation functionality based on
technology used
● Link Adaptation Layer
○ Layer 2.5, takes advantage of layer 2 mechanisms
● Location Service

40. ENHANCING WI-FI BROADCAST FOR V2V
● L2 WiFi broadcast used for all the V2V communications:
● Challenge with IEEE 802.11
○ No collision prevention/ detection/recovery mechanism for
broadcast transmission
● Solution: Wi-Fi broadcast support for VANET
○ Packet forwarding algorithm
■ Assumption: each vehicle is equipped with GPS and Digital
Map
■ Forwarding strategy by spreading NDN Interest packets in all
direction implemented in Link Adaptation Layer

41. ENHANCING WI-FI BROADCAST FOR V2V
● LAL uses:
○ Forwarding timer
○ Computation of timer
1
D(sender,receiver)
where D distance computed using the location service; and a small
random component used to randomize the transmission line
timer=

42. DEMONSTRATIONS

43. EXPERIMENTS
● 10 cars
● Two applications over NDN:
○ Info-traffic:
■ emulates traffic request for a specific location
■ Area encoded in the Interest carried in Interest Packets
■ Name intersections and streets stemming used instead of
numbers
■ i.e./traffic/westwood-at-strathmore/
■ Car from this location can effectively respond to the Interest
○ Road Photo:
■ Represents photo requests from a location
■ Any vehicle that has been to this location can respond

44. EXPERIMENTS
● Vehicular Application Domains
○ V2V
○ I2V (fig 2(b)
○ V2I (fig 2 (c) - 2 (a))
○ Network disruption due to rapidly changing topology and short link duration
○ In-network storage: caching

45. EXPERIMENTS
● Still, platooning, moving around
campus
● Fig. 4(a) shows the CDF for the
number of retransmissions for the
InfoTraffic application in all the 3
types of mobility.
● static case: 75% of the packets need
no more than one retransmission.
● Mobility: this number goes down to
about 65%, however the type of
mobility (either on the P8 roof or on
the roads) has a negligible impact on
the number of retransmissions.
● 95% of the packets are acknowledged
within 5 retransmissions or less (the
max-retransmission was set to 7),

46. EXPERIMENTS: CACHING
ANALYSIS CACHE/FORWARD
STATISTICS
● For consumers & mules
● Caching is more effective
during mobility
● Limited mobility
● Mules observation: 66% of
the Interests were found
using the local cache

47. NDN Operation in multihomed environment
● Two cars (consumer, producer)
● We ran the Road-Photo application: the
consumer requested a photo to be taken by
the producer. Interest and Data packets
were transmitted via all available interfaces.
● Photos were taken in real-time upon
receiving an Interest, their sizes were
between 68KB and 100KB. Each photo was
split into several Data packets of 1300 bytes
each.
● Fig. 5 shows on which interfaces the
consumer received a chunk of content. The
consumer was able to seamlessly receive
consecutive chunks of the same picture
from different interfaces via different
communication channels.

48. V2V COMMUNICATION AT SCALE

49. V-NDN AT SCALE
● Fig 6a & 6c shows that when
the # of cars interested in the
same information increases ,
system performance improves
substantially as measured by
the satisfaction time &
overhead matrix
○ caching and data mules
○ Faster response
● Fig 6b: 35% of Interests are
already acknowledged even
before being transmitted once
○ caching

50. CONCLUSION

51. DISCUSSIONS
● V-NDN removes the isolation between applications and network
transport, allowing forwarding nodes to handle data based on application
needs.
● The communication can start spontaneously due to caching
● Furthermore, locally produced data and data with local meaning, such as
traffic information, no longer need to be transferred to remote servers
before being available to neighbor nodes;
● Data that is produced and consumed in loco can remain in loco and be
delivered to the consumers along the shortest physical path.

52. CHALLENGES

53. Challenges & Future Work
● Study of a V-NDN forwarding strategy to make the best use out of node
multihoming.
● Data naming: shows that encoding geolocation into names can help direct
Interest forwarding for applications using location-based data; however
other types of applications, e.g. fetching today’s news, are unable to make
use of geolocation.
● security and privacy concerns

54. PAPER DISCUSSIONS

55. ● For scalability purpose, broadcasting in a huge network (e.g. the Internet) is
not a good approach. How can CCN handle this problems? Any mechanism
similar to DNS or Content Broker that could be used in CCN? [Pham, Nhat]
● Content naming issues in CCN [Pham, Nhat] [Taesik Gong] [Sungjoon Park]
○ Same name for data
○ Same data but different name
● Ease of updating Naming and routing. Use of SDN for NDN? [Hyunwoo
Choi]
● Caching data packets & interest packets on CCN & its impact on the E2E
principle [Shah]
● Co-existence of CCN with IP networks [Sungjoon Park]
CCN: DESIGN & PERFORMANCE

56. ● Could the breadcrumbs systems cause mobility problems? [Eric]
● Is CCN scalable like IPv6 [ Romain Olivier]
CCN: MOBILITY & SCALABILITY

57. 1. CCN uses content-based security ( digital signatures and encryption) but it still is vulnerable
to DoS attacks. [Hailu Belay] [ Romain Olivier]
○ Hiding legitimate content
○ Flooding Interest packets
2. Drawbacks of using Digital signatures? Any other ways of enforcing security? [ Romain
Olivier]
3. Fake tags on the network [Soowon Kang]
CCN: SECURITY

58. ● Stakeholders willingness to adopt CCN [Hailu Belay]
● Modification of existing systems e.g. search engines for CCN [ Romain Olivier]
CCN: ADOPTION & COMMERCIALIZATION

59. ● Caching content in NDN and propagating stale information [Romain Olivier]
● How to avoid redundant content in the network [Romain Olivier]
● How does V-NDN used forwarding when hosts have multihoming [Shah]
● Normalization problem for content naming [Wonseok]
V-NDN: DESIGN & PERFORMANCE

60. ● Privacy and trust [Hailu Belay]
● Development and integration of high performance cryptographic algorithms [Hailu Belay]
● Security not addressed [several students]
V-NDN: SECURITY

61. ● Killer application[Hailu Belay]
● Data Retention policy and content regulation [Hailu Belay]
● Willingness to cooperate and share content between vehicles [Hyunwoo Choi]
V-NDN: ADOPTION & COMMERCIALIZATION

62. References
1. Jacobson, V. .et al (2009). Networking Named Content
2. Grassi, G (2014). VANET via Named Data Networking
3. https://named-data.net/project/faq/
4.

63. ROUTING
Figure 2 shows a basic routing scheme in CCN.
1. The client 1 requests content to CCN router H. When CCN router H
receives client 1’s interest packet, it checks its content cache table to
find whether the requested content is in the table or not. If
requested content is found within the cache table, CCN router H
sends the requested content to client 1. However if the content is not
in the cache table, CCN Router H sends an interest packet to other
CCN routers. In this way, each interest packet is sent to the CCN
Router A which has the requested content.
2. CCN router A receives an interest packet from CCN router B and
checks its cache table. Then CCN router A sends the requested
content using reverse path to router H and when each CCN router
receives the contents, it stores the contents into content cache.
Finally, client 1 receives the requested content from CCN router H.
3. The client 2 requests same content which is requested by client 1.
CCN router I receives an interest packet. However CCN router I
doesn’t have the requested content in its cache table. In this case,
client 2’s request message is sent to node D.
4. When node D receives the interest packet, it sends a data packet
including requested content to client 2.