Transcript PPT

COM594: Mobile Technology
Lecture Week 2
Building The Mobile Internet
Internet Sessions
3
A Common Scenario ?






You are cycling to University
Talking on your mobile
You pass many ‘cell-towers’ on the journey
Each time you move between cell-towers you lose the call.
You have to stop and make the call again
You think..’This is a really cool service’..?
4
A Common Scenario ??
 On long journeys you need to contact a new cell-phone provider
and sign up for a temporary contract.
 When you are on an international train and streaming a movie, it
stops periodically and you have to contact a new company and
arrange a new contract before continuing with the movie.
 Then the movie starts over from the beginning…
5
Sessions




I don’t think this would be widely accepted
What users expect (even demand) is seamless real-time mobility.
This has become known as
‘Session-mobility’
The Internet was not designed to enable the implementation of this
kind of provision.
6
The Internet and the Mobile-Internet
 Things get even more challenging when moving through differing
types of access network. e.g. from Wi Fi to Cellular, or between
different operators.
 What we need is a ‘New Internet’
 A ‘Mobile-Internet’
7
The Mobile-Internet
 The Mobile-Internet must be a Pervasive IP-based network that
can link fixed and mobile nodes,
 Nodes, can have many characteristics and types
8
The Mobile Internet








Nodes may be:Sensors;
Servers;
Stand-alone or Distributed;
Battery-powered or Mains Powered;
User initiated or Self Initiating
Permanent or Temporary
State-of-the-Art, or Legacy
9
The Internet: Review of Characteristics
 Lets revisit the Internet and see if we can identify difficulties in
making session-mobility possible:
10
Computers Connected to the
Internet are called Hosts
Routers are special hosts,
that are connected to more
than one network, and that
transmit data from one
network to another. Hosts
communicate using the
TCP/IP protocol suite
Network
Routers
Networks
Network
Network
Network
Hosts
Network
Figure 2-1: The Internet: A Network of Networks
This infrastructure supports much of the current mobile communication but it was
not designed for that purpose. Our objective is to highlight design weaknesses
and consider what solutions might be possible
11
Packet Vs Circuit Switching
Circuit switching is used by traditional telephone circuits  Circuit dedicated to only 1 end-to-end
communication for the duration of the session  This justified Packet Switching using TCP/IP
The key benefit of TCP/IP is that it enables multiple communication sessions over the same
communications channel (Multiplexing). A problem is that the order of the packets is not necessarily
preserved and so re-ordering is required at the receiver
Network
Original data
Packetized data
Re-assembled original data
Figure 2-2: Packet-switching: Dividing a data Source into Packets at the Sender,
and Reassembling the Data-Source at the Receiver
12
IP Over Everything. Everything over IP
The key concept in the implementation of TCP/IP is Internet Protocol (IP) that is used to
transport data bits from source to destination.
IP shields the underlying network technology from the applications that run on the network. It
also makes the physical network invisible to the applications.
Web
e-Mail
Telnet
Application
WiFi
Data Link
IP
Ethernet
DWDM
13
Addresses
For true IP networking, and Internetworking to be possible every host on the
Internet must be associated with a globally-unique IP address
IPv4 has a 32-bit wide address. This means that there are exactly 232 possible
IP addresses available. It has become clear that this is not enough.
IPv4 has been superseded by IPv6. This has a 128 bit word. So there are 2128
unique IP addresses in this scheme.
Do you think this will be enough? Think about the potential demand generated
by the ‘Internet of Things’.
Can you see any problems with the implementation of the IPv6 concept?
14
A problem with IP addressing is that an IP address has two functions.
Firstly it is used to identify a host. Secondly, and at the same time, it is
used to identify the location of the host.
As it turns out, this simple design decision made some 40 years ago, is
a serious problem for mobility
Bit
0
0
4
8
Net-id
12
16
20
24
28
Host-id
Figure 2-4: IP Addresses consist of a net-id and a host-id part.
Same for all hosts in a
particular network
32
15
In the initial implementation of the Internet IP addresses were split at a
predefined and fixed boundary. In this system networks were forced into rigid
Classes
For example, Class A networks had the first 8-bits of the IP address represent
the net-ID and the remaining 24-bits available for hosts. This meant that a
Class A network could have 2^24 hosts, or 16,777,216 hosts. The net ID
determined that 2^8 = 256 such networks could exist.
A class B network had a 16/16 split and so 2^16= 65,536 networks each with
65,536 hosts could be implemented
Class C networks had 24-bits for the net-id with 8-bits for the hosts. So
16,777,216 networks were available with 256 hosts each
16
Classless Interdomain Routing (CIDR) Subnetting
10.0.0.0 / 8
10.1.0.0 / 16 10.2.0.0 / 16
Class C
10.3.0.0 / 16
Class A
Class B
10.1.0.0 / 24 10.1.1.0 / 24 10.1.2.0 / 24
10.1.2.0 / 28
10.1.2.128 / 28
Subnetting: A Large network can be divided into smaller subnetworks
Aggregation: 2 or more networks can be combined into a larger network
17
The IPv4 Packet Format
Bit
0
0
4
Version
The IPv4 packet
consists of a host IP
address, a destination
IP address, the payload
containing the
application data and
additional fields to
support routing,
transportation and errorchecking functions.
8
IHL
16
12
20
Type of Service
24
28
Total Length
31
Identification
Flags
Fragment Offset
63
Time to Live
Protocol
95
IPv4 Source Address
127
IPv4 Destination Address
Header Checksum
159
Options
Padding
191
Payload (variable length)
31
18
IPv6
 New version of the Internet Protocol defined by the
Internet Engineering Task Force due to the foreseen
shortage of IPv4 address space.
 It has a much larger address space by using 128-bit
instead of 32.
 IPv6 contains other features to automatically assign
addresses to hosts and to make routing more efficient
Bit
0
32
64
0
4
Version
8
12
16
20
Traffic Class
24
28
32
Flow Label
Next Header
Payload Length
Hop Limit
IPv6 Source Address
128
192
IPv6 Destination Address
256
320
Payload (variable length)
Figure 2-7
IPv6 Packet Format
Documented in RFC 2460
20
Network Address Translation
 A cornerstone of Internet implementation is that every host should
have a unique IP address.
 Currently, this is not strictly the case;
 Networks use Internal IP addressing;
 Not exposed to the Internet
21
Network Address Translation
 Perceived Benefits?
 Security: IP address of specific user hidden!
 Easier Network set-up with large address ranges ‘internal’ to the
local network.
 Reduces demand for Global IP addresses
 (Won’t be needed with IPv6)
22
Network Address Translation







Disadvantages:
Internal and External IP addresses must be unique
Special range of IP addresses reserved for Internal use:
10.0.0.0/8
172.16.0.0/12
192.168.0.0/16
Translation Process required (NAT)
23
NAT
 Communication across the Internet requires interaction between
globally unique IP addresses
 A host with a locally-unique IP address communicating with a remote
host, requires IP address translation at a host which has a globallyunique address.
 The translation process (NAT) breaks the normal communication
between Internet hosts.
24
TCP/IP 5-Layer Model
 The Layered structure of the TCP/IP ‘Stack’ has been the
prime reason for its success.
 Each layer is independent of the other layers. Each layer
receives a service from the layer immediately below it,
and provides a service to the layer above it.
25
TCP/IP 5-Layer Model
 The data in any one layer, is carried as a payload in the ‘packets’
of the layer below it. (this is called Encapsulation)
 Data in the higher layers is ‘largely opaque’ to the layers below it.
 The phrase largely opaque recognizes the fact that it is
sometimes useful to ‘leak’ information from one layer to another.
This strictly is in violation of the layered-principle.
26
Application Layer
Application Layer
Transport Layer
Transport Layer
Internet Layer
Internet Layer
Data Link Layer
Data Link Layer
Higher
Layers
Lower
Layers
Physical Layer
Figure 2-8 The TCP/IP Five-Layer Model
27
Data Link Layer
Internet Layer
Transport Layer
Application Layer
Application data
Figure 2-9 Data from Higher Layers Encapsulated Lower-layer Packets
28
Layer 1: Physical Layer
 Not strictly speaking part of the TCP/IP model.
 This layer carries the actual data over the transmission medium.
 For each hardware technology , an addressing mechanism must
be defined to transport information over LANs or WANs. These
hardware addresses must be unique inside a specific network and
for a specific technology, but not necessarily globally (IP takes
care of global comms)
29
Layer 2: Data Link Layer
 Also called the network interface layer, is responsible for
accepting IP packets from the Internet Layer (above) and
encapsulating them in a protocol that is specific for the
Layer 1 network technology
 Examples of Protocols:
Ethernet, Address Resolution Protocol (ARP), Layer 2 Tunneling
Protocol (L2TP), Point to Point(PPP), Digital Subscriber Line
(DSL), Integrated Services Digital Networks (ISDN), etc.
30
Ethernet (IEEE802.3)
• Ethernet (IEE 802.3) implements a unique MAC
address to the physical interface card
• The MAC address is unique,
• Specified by the hardware manufacturer
• Cannot be used for Interdomain routing
31
Address Resolution Protocol
 To map an IP address to a hardware address requires an Address
Resolution Protocol (ARP).
 Typically a host receiving an IP address broadcasts a message to
all the hosts to which it is connected asking for the mac address
that the IP address is associated with.
 The specific host responds with its MAC address.
32
ARP and Caching
 In an effort to improve efficiency, host routers cache MAC-IP
associations
 If a host changes its IP address it could respond with its cached
address rather than its new address.
33
Layer 3: Internet Layer
 Responsible for handling the communication between hosts on the
Internet. (It performs a routing function)
 It accepts a packet from the transport layer, turns it into an IP
packet and hands it to Layer 2 (and vice versa)
34
Layer 4: Transport Layer
 Responsible for providing communication between application on
different hosts.
 Takes the data stream that comes from a program and divides it
into packets.
 It includes information on what application transmits the data and
which application at the receiving host is supposed to receive it 
it uses an abstract endpoint of a communication: the protocol port
35
Layer 4: Transport Layer
 Examples of Ports:
TCP Port 80: used for HTTP
TCP Port 25: used for Simple Mail Transfer Protocol (SMTP)
 Examples of Transport Layer Protocols:
Datagram Congestion Control Protocol (DCCP), Stream Control
Transmission Protocol (SCTP), Transmission Control Protocol
(TCP) and User Datagram Protocol (UDP)
36
UDP Datagram
UDP is the main example of a transport protocol that
is unreliable:
Bit
At the transport level the receiver host will not
acknowledge receipt of packets and there are no
measures to limit (for example) the data rate at which
packets are sent in order to take the available
bandwidth into consideration.
If reliable connections are needed, they will have to be
implemented by the application.
The main advantage of UDP is that the protocol can
be very simple and therefor efficient and fast
0
0
4
8
12
16
20 24
Source Port
Destination Port
32
Length
Checksum
64
Payload (variable length)
28
32
37
TCP Packet
This is the most reliable transport protocol.
TCP relieves application programmers of the burden
to have to implement delivery confirmation and error
detection
Bit
0
0
4
8
12
16
20 24
Source Port
28
Destination Port
32
Sequence Number
64
TCP Will:
• Make sure that the sender is informed if a packet
doesn’t reach its destination
• Adapt the rate of sending packets to the available
bandwidth
• Correct errors in transport (by asking for packet
re-transmissions)
Acknowledgment Number
96
128
160
Offset
Reserved
Checksum
Flags
Window Size
Header Checksum
Urgent Pointer
Options (variable length 0-320 bits,
depending on offset field)
Payload (variable length)
32
38
Layer 5: Application Layer
 Layer at which the user application exists and that uses TCP/IP
networks to communicate over.
 The Applications interact with one of the protocols in the transport
layer to send and receive data
39
Layer 5: Application Layer
 Examples of Application Layer Protocols:
File Transfer Protocol (FTP), Network Time Protocol (NTP), Post
Office Protocol (POP) Internet Message Access Protocol (IMAP),
Session Initiation Protocol (SIP), Simple Mail Transfer Protocol
(SMTP), Simple Object Access Protocol (SOAP), Secure Shell
(SSH), Dynamic Host Configuration Protocol (DHCP), Domain
Name System (DNS), Hypertext Transfer Protocol (HTTP) and
Telnet.
40
Socket API
 Some kind of Application Programming Interface (API) is
needed to create an application that connects to another
application on the Internet.
 The API needs to be capable of creating connections to
other hosts or wait for incoming connections from other
hosts.
 Most operating systems have implemented an
abstraction of an endpoint for a communication session
called a socket
41
Socket API
 The O/S or the program itself associates a socket identifier with the
IP address of the host and a port.
 If it wants to connect to a remote host, it has to associate a remote
address and port with the socket identifier as well
 This process is called binding.
Sockets, Protocols and Ports
Application 1
Application 2
Application 3
Application 4
Socket
references
TCP
Sockets
TCP
Ports
Sockets
bound
to ports
1
2
3
1
65535
2
3
65535
UDP
TCP
IP
Sockets Protocols and Ports
43
Sessions and Mobility
 A TCP communication is identified with the 5 tuple:
Local IP Address
Local Port
Remote IP Address
Remote Port
Socket Identifier
44
Sessions and Mobility
Use of Socket API
 An important consequence of the way that applications use TCP/IP
through the socket API:
If the IP address of a host changes, the identifier for the session
must also change
45
Sessions and Mobility
 The entire structure was built on the implicit assumption that IP
addresses were essentially fixed and that Hosts did not change
location.
 If a host moves they need a new IP address;
 If a host gets a new IP address then the Socket ID for the
communication must change
 Therefore the session breaks
46
Sessions and Mobility
 Another consequence of the TCP/IP inherent design is that an IP
address in fact performs two functions;
 It is a location-identifier
 It is an end-point-identifier
 In other words, an IP address identifies the specific host, and also
its location in the network.
 Not an issue until hosts start to move around…!
47
Building the Mobile Internet
 Building the mobile internet is really about finding solutions for the
fact that communication sessions are tied to a specific transport
session and thus a set of IP addresses.
 A number of strategies can be followed:
48
Strategies





Accept that application sessions will break;
Introduce an application layer ‘session-persistence’ mechanism
Keep the same IP address when moving;
Introduce a new Layer
Redesign the TCP/IP protocol stack to achieve separation of
locators and end-point identifiers
49
Accept that Application Sessions will
Break
 Easy option
 For some situations session breaks don’t matter (i.e. receiving your
email).
 Nomadic mobility: Focus on seamless access to different networks
(does not require session persistence)
 User can switch networks seamlessly
 However, sessions will not necessarily ‘stay up’
50
Introduce a ‘Session-Persistence’
mechanism in the Application Layer
 Requires the Application to implement a session state not bound to
the transport layer;
 Probably impractical
 Maybe not desirable
Cookies in web browsers do this currently
Big security and privacy issues
51
Keep the same IP Address when Moving
 The IP layer is an abstraction layer on top of the physical
infrastructure,
 When a host moves, it must change its IP address so that a source
knows where to route its packets
 This can be done to an extent by ‘tunneling’.
 Make multiple physical layers look like one layer to the IP layer by
encapsulating layer 2 packets in other layer 2 packets
52
Introduce a New Layer
 Consider implementing a session layer between the TCP layer and
the Application Layer.
 This is akin to revisiting the ISO-OSI Seven layer model which had
a Session Layer above the TCP layer and under the Application
layer.
 Perhaps we used the wrong stack after all…?
53
Redesign TCP/IP to Separate Location ID
and Host ID
 A fairly fundamental approach
 Separate entities to describe
A node
Its location
 Possible in theory
 Difficult to implement
 Maybe the only realistic way forward
 Explored in some detail later