Sample

1
Mobile Computing Gihwan Choghcho_at_cs.chonbuk
.ac.kr
2
Content

• Getting start
• distributed computing, what means for MC?
• Internet protocol, and its considerations for MC
• Next generation Internet
• Cellular technology overview
• Internet host mobility
• routing optimization

3
Distributed Computing (I)

• Whats distributed computing? by Donovan
• A computing paradigm which is provided by a
  collection of computers connected by a
  communications subnet and logically integrated in
  varying degrees by a distributed operating system
  and/or distributed database system on purpose to
  resolve a task co-operatively
• Each node has autonomous mechanisms which also
  coordinate their operations through a global
  mechanism
• Background
• price vs. performance revolution in computer
  hardware
• cost effective and efficient communication subnets

4
Distributed Computing (II)

• Why distributed computing? by Donovan
• distributed nature of real world
• equipment cost
• user know-how and control
• flexibility and configurability
• etc.
• Technical Aims
• resource sharing
• location transparency
• Models
• system client-server
• programming RPC (Remote Procedure Call), CORBA

5
Distributed Computing (III)

• Transparencies
• location, execution, device, program code ...
• Advantages
• good performance, reliability, resource sharing
  and extensibility...
• Application Spectrum
• e-mail ... command control ... resource sharing
  ...
• Using Fields
• banking, university computing, factory, office
  automation ...

6
OS Perspectives

• NOS (Network OS) vs. DOS (Distributed OS)

Host 1
Host 2
Host 3
Host 4
DOS Examples V system, Eden, Amoeba Locus,
Mach, Spring...
Mach
Mach
Mach
Mach
7
Multiprocessor System vs. Distributed System

• Tightly-coupled vs. Loosely-coupled

A Computer
Processor
Processor
Shared I/O Devices
Shared Memory
Network
8
Why should be Distributed Algorithms changed?

• Underlying network structure changes with hosts
  moving
• no more logical structure of infrastructure can
  be utilized
• whenever an algorithm tries to refer overall
  structure, a physical structure which reflects
  current situation should be repeatedly
  reconstructed, and then convert it to a logical
  structure
• communication costs are no longer the same for
  all hosts in a logical subnet (source host would
  be still keep moving)
• Mobile hosts may disconnect or doze off!
• offload and/or download
• Broadcast communication is available for a group
  of hosts using only one transmission

9
Ex) Distributed Mutual Exclusion

• Lamports Algorithm
• each host maintains a logical clock
• each host maintains a request queue that contains
  messages in increasing order of timestamps
• host hj requires a resource by sending a
  timestamped request message to all others and
  inserts it into its queue. Each host which gets
  this send a timestamped reply message, and
  inserts this request into its queue
• hj can access the resource when its request is at
  the head of the queue, and it has received
  replies from all other hosts with higher
  timestamps
• when it is done with the resource, it sends a
  release message to all others. The recipients
  delete the request from their queues

10
Ex) Distributed Mutual Exclusion (Suppl. -1)

• Enters critical region

8
0, 8
2,12
0
0
0, 8
8
12
OK
OK
OK
8
2
2
1
1
12
OK
0, 8
0, 8
2,12
2,12
2,12
2,12
12
Enters critical region
2,12
Make a Decision
host id, timestamp
11
What problems and what should be changed?

• Problems
• high search costs for all MH - MH messages
• all hosts required to participate with running
  the algorithm, so no doze or disconnect can be
  allowed
• data structures need to be maintained at the MHs
  (with lots of messages exchanged), so higher
  power usage
• The algorithm be possibly changed as
• only MA participates in the coordination
  (Indirect model). An MA treats all requests from
  MHs within its cell as if they were its own
  requests
• an MH simply initiates the process by sending a
  init_req() message to the MA. The MA then
  processes, waits and maintains the request on
  behalf of the MH
• when this virtual request is ready for execution,
  the MA sends a grant_req() message to the MH
• the MH sends a rel_req() when it is done, the MA
  then sends a release message to all others

12
The Internet (I)

• The collection of networks and gateways that use
  the TCP/IP protocol suite and function as a
  single, cooperative virtual network
• Virtual Circuit
• connection set up at the beginning
• connection remains throughout
• reliable Communications
• ex) TCP
• Datagram
• each datagram routed separately
• each contains an address
• no guarantee of delivery
• ex) IP and UDP

13
The Internet (II)

• The TCP/IP Stack
• Well-known TCP/IP services
• mail (SMTP), ftp, rlogin, telnet, rcp, X window
  clients ...
• rpc, rfs, nfs, rwho ...

Host A
Host B
Uses TCP/IP Services
Application
Application
TCP
Virtual Circuit
TCP
Gateway G
Routes Datagrams
IP
IP
14
Internetworking with TCP/IP
15
IP(Internet Protocol)

• A user thinks of an internet as a single virtual
  network that interconnects all hosts, and through
  which communication is possible its underlying
  hardware is both hidden and irrelevant
• IP provides three important definitions
• defines the basic unit of data transfer
• performs the routing function
• includes a set of rules that embody the idea of
  unreliable packet delivery, such as packet
  processing, error control
• Connectionless delivery service
• unreliable, best-effort, connectionless

16
Internet Addresses (I)

• Universal communication service requires a
  globally accepted method of identifying each
  computer that attaches to it
• Three host identifiers
• names what object is (a location independent
  characteristic of a network entity)
• addresses where it is (a function of the
  location of the destination)
• routes how to get there (something that depends
  on both the source and destination)
• Internet address was made to standardize on
  compact, binary addresses that make computations
  such as the selection of a route efficient

17
Internet Addresses (II)

• Each host on a TCP/IP internet is assigned a
  unique 32-bit internet address that is used in
  all communication with that host
• IP address is a pair (netid, hostid), where netid
  identifies a network, and hostid identifies a
  host on that network
• Because IP addresses encode both a network and a
  host on that network, they do not specify an
  individual computer, but a connection to a network

18
Internet Address (III)

• consists of four Octets (8 bit chunks)

0 8 16
24 31
Class A 0 networkid hostid
0-127...
Class B 10 networkid
hostid
128-191...
Class C 110 networkid
hostid
192-223...
Class D 1110 multicast address
224-239...
Class E 11110 reserved for future
use
240-254...
hostid (or all bits) set to 1 broadcast
networkid (or hostid) set to 0 means this
19
IP Address (Considerations)

• With current IP address,
• IP datagrams clearly are routed based on the
  networkid (and subnetid) in the IP address
• networkid (and subnetid) do not specify an
  individual machine, but a connection to a network
• all hosts having addresses with the same
  networkid (and subnetid) are connected to the
  same physical network
• When we consider host mobility,
• If a host moves from one network to another, its
  IP address must change. - D. E. Comer,
  Internetworking with TCP/IP, pp 65
• 4.8
  weaknesses in Internet Addressing
• how it can be resolved?
• in addition, It may need another facility to
  trace the physical location of moving host in
  order to find out the hosts current location

20
Internet Routing

• Note the hierarchical structure of the IP
  address aims to scale internetworking well by
  providing the abstraction of address clustering.
  This implies to the routing domain as
• allows routers to keep minimal routing
  information, so a gateway needs to maintain
  networkid, not full IP address only specifies one
  step along the path to a destination network
• make their routing decisions efficiently by
  allowing a gateway to use default routes to
  possible distant destinations
• permits some degree of autonomy to organizations
  to do their own internal routing structure (with
  subnetting)
• Table-driven IP routing
• Basic principles
• each router announces which networks it can reach
• each router remembers the others announcements
• remembered routes time out to delete stale routes

21
Routing Algorithm

• Route_IP_Datagram(datagram, routing_table)
• Extract destination IP address, ID, from datagram
  Compute IP address
• of destination network, IN
• if IN matches any directly connected network
  address
• send datagram to destination over that network
• else if ID appears as a host-specific route
• route datagram as specified in the table
• else if IN appears in routing table
• route datagram as specified in the table
• else if a default route has been specified
• route datagram to the default gateway
• else declare a routing error

22
Internet Routing (Considerations)

• The Internet topology is rapidly growing or
  changing
• In this situation, in order to provide an
  Internet-wide routing service, the Internet uses
  an architectural approach that allows groups to
  manage local gateway autonomously, adding new
  network interconnections and routes without
  changing distant gateways
• In the mobile computing model, a packet routing
  path bound for a mobile host is mainly decided on
  the fixed network
• Current IP routing mechanisms cannot decouple the
  host tracing function from an IP address
• The Internet protocol needs another facility to
  trace a moving host, and then to deliver the
  packets following the host based on the location
  information

23
Which the Internet layer should take charge of
host mobility? (I)

• If applications do this, each time an application
  wants to communicate with another, it must obtain
  the current address of its peer. Moreover, it is
  impossible to provide on-line moving without
  modifying the application program itself, so as
  to re-establish the existing session with the new
  location
• If the transport layer do this, it does not see
  the notion of host with the same reason as above.
  Thus, host mobility causes increased delays and
  packet losses. TCP interprets these as signs of
  network congestion - so, it throttle its
  transmissions, further degrading performance

24
Which the Internet layer should take charge of
host mobility? (II)

• The internet layer hides the different hardware
  address, and resolves the exact location of a
  host on an internetworking its own addressing and
  routing facilities. Therefore, it could give to
  higher level protocols the abstraction that the
  network address remains unchanged
• If the network interface layer do this, there is
  no way to maintain host location in the network
  (with physical address). Also, a bridge-based
  routing scheme is not enough (too restricted).
  However, it could have some useful features, such
  as monitoring connection/disconnection to/from
  communication medium, then reporting it to higher
  layer

25
The Internet Protocol (Considerations, I)

• When the source host specifies the destination
  host's address, a binding between the address and
  its route is established, and thereafter no
  re-evaluation of the binding takes place in a
  static network, there is little re-evaluation of
  the binding
• In a mobile computing environment, mobile hosts
  are expected to move from time to time, and in a
  way that necessitates changing their address -
  moves to other locations in different parts of
  the hierarchy
• In order to re-evaluate of the binding at times,
  the IP-based approach for supporting host
  mobility can be formalized as a location problem

26
The Internet Protocol (Considerations, II)

• The location domain includes an addressing
  convention for identifying mobile hosts, and
  acquisition and/or preservation of locale
  information
• In addition, it is inevitable that the routing
  scheme must be re-structured as well, according
  to the address convention adopted
• The location issue seems similar with the mapping
  elaboration between name and address, that is,
  the mapping what to where is changed to the one
  known-where to current-where

27
Three Addressing Schemes for host mobility

• PAS (Permanent IP-Address Scheme)
• each MH has a permanent IP address from the
  initial (home) administration address space
• whenever an MH moves, some hosts or routers (at
  least the home MA) are informed of the current
  MAs address
• the hosts or routers forward packets, which are
  passing through them, to the current MA using the
  location information recorded

Internet
Packets from the Internet
MA p
MA s
MA k
MH k, MA s
MH k,
MH k
Host Moving
28
Three Addressing Schemes (Cont.)

• TAS (Temporary IP-Address Scheme)
• a temporary address is assigned dynamically every
  time a host connects with an MA
• the location information is managed by supporting
  a directory or the source host broadcasts a query
  to find the current location
• the source host directly routes packets using the
  location information obtained

Who serves MH k at a moment?
Internet
Thats me!
MA k
MA p
MA s
MH k, -
MH k, MH t
MH a
MH k
Host Moving
29
Three Addressing Schemes (Cont.)

• ENS (Embedded Network Scheme)
• each host has a permanent IP-add. and an embedded
  network add. which consists of the current MAs
  add. and a temp. add.
• the gateways maintain a mapping between
  IP-addresses and the embedded network address
• the gateways use the mapping to forward any
  on-going packets

Notify MH ks address to the Internet
Internet
MA p
MA s
MA k
MH k, (MA s,MH t)
MH k, (MA s,MH t)
MH k, (MA s,MH t)
MH a
MH k
Host Moving
30
Location Considerations

• A routing decision must be made based on the
  location information that is available - packet
  routing efficiency depends critically on how
  effectively a packet comes across its current
  address
• The packet routing paths that go with host
  mobility depend decisively on the somewhere which
  holds the information for a mobile host's
  physical locator
• A trade-off in devising a location scheme
  excessive location preservation can be wasteful
  of network resource, whilst insufficient location
  propagation leads to inefficient routing

31
Common Approach

• Addressing uses two IP addresses, that is, a
  separation the dual nature of an IP address into
  a logical identifier which is the permanent
  (home) IP address of the host, and a physical
  locator which is a forwarding (current) IP
  address
• Routing uses tunneling technique, that is, a
  forwarding mechanism in association with location
  caches held around the network
• The arguments that host mobility support system
  now faces are
• how to distribute location information,
• and then how to utilize the information
  effectively, in order efficiently to deliver
  packets to moving destinations whilst still
  limiting costly location updates as much as
  possible

32
Content

• Getting start
• distributed computing, what means for MC?
• Internet protocol, and its considerations for MC
• Next generation Internet
• Cellular technology overview
• Internet host mobility
• routing optimization

33
The Internet Protocol

• Internet is not a physical network, but it is a
  method of internetworking physical networks and a
  set of conventions for using networks that allow
  the computers they reach to Internet
• The collection of networks and gateways that use
  the TCP/IP protocol suite and that function as a
  single, cooperative virtual network
• Network level interconnection scheme
  (as opposed to application level
  interconnection)
• connectionless packet delivery service
• reliable stream transport service

34
National Information Infrastructure

• "The NII will provide all Americans with the
  information they need, when they need it and
  where they need it, at an affordable cost."
• Dr. Jack Gibbons, Presidential Science Advisor
• ARPA HPCC Symposium, 15 March 1994
• "Hundreds of different networks, run by
  different companies and using different
  technologies, all connected together in a giant
  'network of networks,' providing telephone and
  interactive digital video to almost every
  American."
• Vice President Al Gore, Jr.
• Address to the ITU, 21 March 1994

35
Next Generation Internet (NGI) (I)

• The NGI initiative is a multi-agency federal
  research and development program, that aims to
• http//www.ngi.gov
• new technologies and services sponsor research
  and development in new networking technologies
  and services in support of the high performance
  applications requirements
• testbed(s) build a high performance network
  infrastructure (100 to 1,000 times faster
  end-to-end than today's Internet) in support of
  both network research and science applications
  research
• applications support demonstration of next
  generation applications requiring advanced
  networking technologies
• It began October 1, 1997, with the following
  participating agencies DARPA, NASA, NIH, NIST,
  NSF, (DE)

36
(No Transcript)
37
Internet 2

• UCAID (University Corporation for Advanced
  Internet Development) is supported by over 175
  member organizations. universities, corporations
  have joined to advance networking in higher
  education
• http//www.ucaid.edu
• Internet 2 is a collaborative project by over 120
  U.S. research universities engaged in the major
  challenges facing the next generation of
  university networks
• http//www.internet2.edu
• Abilene is a project to develop a nationwide
  advanced network to serve as backbone network for
  the Internet2
• http//www.ucaid.edu/html/abilene.html

38
(No Transcript)
39
IP Next Generation (IPng)

• IPng is an IETF WG intended to provide IPv6 which
  is designed to be an evolutionary step from IPv4
• http//www.ietf.org/html.charters/ipngwg-charter.
  html
• http//playground.sun.com/pub/ipng/html/ipng-main
  .html
• Its motivations are
• limited number of available addresses
• difficulty in managing routing tables
• need to support high performance network (e.g.
  ATM), at the same time, low bandwidth network
  (e.g. wireless)
• 6Bone is the IPv6 backbone that was set up to
  assist in the evolution of IPv6 in the Internet
  
• http//www.6bone.net/

40
6Bone
41
IP version 6 (I)

• IPv6 is the formal name of the protocol
  recommended by the IETF IPng group, its
  objectives are
• The Recommendation for the IP Next Generation
  Protocol, RFC 1752, Jan., 1995
• Internet Protocol, Version 6 (IPv6)
  Specification, Internet Draft, Nov., 1997
• support large global internetwork
• support new low-end Internet devices
• (PDAs, mobile computers, consumers,
  devices)
• support the networked multimedia services
• Implementations
• Apple, BSDI,Bull, Dassault, Digital, Epilogue,
  FTP Software, IBM, INRIA, Linux, Mentat,
  Microsoft, Novell, NRL, NTHU, Pacific Softworks,
  Process Software, SICS, SCO, Siemens Nixdorf,
  Silicon Graphics, Sun, UNH and WIDE

42
The Challenges from IPv4

• Plenty of addresses
• Reduced administrative overhead
• Opportunity for better routing
• Support for address renumbering
• Improved header processing
• Reasonable security
• Support for host mobility
• QoS control capability

43
IPv6 Design (I) Addressing

• Two-level structure of the IPv4 address, what?
• Address paces are
• 340,282,366,920,938,463,463,374,607,431,768,211,4
  56 (296 times that of IPv4)
• An address is represented as xxxxxxxx (x
  is 16 bit long) (ex, fedcba4500d44354f345ad23
  546d232c)
• compression 0s (ex, ff0100000043 gt
  ff0143)
• combination between the IPv4 address and IPv6s
  one
• IPv4 compatible address gt IPv4 address
• (eg. xxxxxxd.d.d.d)
• IPv4 mapped address gt ffffIPv4 address
• IPv6 addresses are identifiers for interfaces,
  not nodes
• A single interface may be assigned multiple IPv6
  addresses of any type, that is, unicast, anycast,
  multicast

44
IPv6 Addressing (Cont)

• Unicast
• provider based address 1/8 fraction of address
  space
• link (or site) local use address 1/1024 fraction
• anycast use unicast address format
• multicast

010
REGISTRY
INTERFACE
SUBNET
SUBSCRIBER
PROVIDER
(4)
(4)
Usually IEEE802 48 bit address
11111111
GROUP ID
FLGS
SCOP
45
Whats in an IPv6 Datagram
15
31
0
Vers
Prior
Flow Level
Payload Length
Next Header
Hop Limit
Source Address (128)
10 X 32 bit 40 octets
Destination Address (128)
Next Header
Header Length
Hop-by-hop option (variable)
Next Header
Header Length
Other option headers
IP payload TCP header (variable)
46
IPv6 Design (II) Performance

• To meet performance requirement on the NGI
• reduce the number of fields in the datagram
  options are placed in separate optional headers,
  and most of these optional header are not
  examined on in-between routers
• fix the length of header IPv6 extension headers
  act as a separated extension headers with
  arbitrary length
• packet fragmentation is not performed by IPv6
  routers, but by the source host only

47
IPv6 Header Extensions

• Currently defined extension headers (in sequence)
• hop-by-hop hop by hop processing on the router
• routing similar to the source record route for
  IPv4
• fragment fragmentation / reassembly
• authentication packet integrity and
  authentication
• encapsulating privacy
• destination processed at the final destination
  only
• - jumbo payload option 232 octets

48
IPv6 Design (III) QoS Capabilities

• QoS is controlled by the flow label and the
  priority field
• Priority (4bit)
• congestion controlled traffic (0 7), such as
  back-off
• internet control traffic snmp
• interactive traffic on-line user-to-host
• attended bulk traffic ftp, http
• unattended data transfer email
• filler traffic USENET
• uncharacterized no priority
• non-congestion controlled traffic (8 15)
  constant (at least smooth) data rate and delivery
  delay from most willing to discard(8) to least
  willing to discard(15)

49
IPv6 QoS Capabilities (Cont)

• Flow is a sequence of packets sent from a source
  to a destination a flow is uniquely identified
  by the combination of a source address and a
  24-bit flow level
• Flow label is used by a source to label a flow
  for which it requests special handling by the
  intervening IPv6 routers, such as real-time
  service
• the nature of special handling might be conveyed
  to the routers by a control protocol, such as a
  resource reservation protocol, before the source
  start to send
• a router can decide how to route and process
  these packet by simply looking up the flow label
  in a table, without examining the rest of the
  header
• the flow level is chosen randomly and uniformly

50
IPv6 Design (IV) Routing

• Almost identical to IPv4, but new routing
  functionality
• provider selection (based on policy, performance,
  cost)
• host mobility (route to current location)
• auto-readdressing (route to new address)
• These functionalities are achieved by creating
  sequences of IPv6 addresses using ipng routing
  option, which is very similar to IPv4s LSRR
  option (cf. home-based tunneling)

51
IPv6 Design (V) Security

• Application specific security mechanism on IPv4,
  such as privacy enhanced mail, secure http, what
  problems?
• An Overview of a security architecture, RFC
  1825, Aug., 1995
• Description of a packet authentication extension
  to IP, RFC 1826, Aug., 1995
• IP level security could ensure the
  interoperability between the secured packet and
  unsecured packet
• it has two functional areas authentication and
  privacy
• Support for security features could be
  implemented both, but mandatory for IPv6 and
  optional for IPv4

52
IPv6 Transition

• ngtrans is an IETF WG which is responsible for
  the transition of the Internet from IPv4 to IPv6
  
• http//www.ietf.org/html.charters/ngtrans-charter
  .html
• Aims to allow IPv6 and IPv4 hosts to interoperate
• incremental upgrade and deployment (one by one
  installation)
• minimal upgrade dependencies (DNS only for IPv6
  address record)
• easy addressing (inter-use two types addresses)

53
IPv6 Transition (Cont)

• Step 1 IPv4 gtgt IPv6 (number of hosts)
• dual stack model
• Step 2 IPv4 IPv6
• tunneling IPv6 packet within IPv4 header
• Step 3 IPv4 ltlt IPv6
• header translation

Application
TCP, UDP
IPv6
IPv4
Ethernet, FDDI, etc.
54
21 Century Applications

• Enabling applications
• collaboration technologies
• digital libraries
• distributed computing
• privacy and security
• remote operation and simulation
• Disciplinary applications
• basic science
• crisis management
• education
• the environment
• federal information services
• health care
• manufacturing

55
21 Century Applications (Cont)
56
Content

• Getting start
• distributed computing, what means for MC?
• Internet protocol, and its considerations for MC
• Next generation Internet
• Cellular technology overview
• Internet host mobility
• routing optimization

57
The Vision
The Vision Revisited "It is dangerous to put
limits on wireless." Guglielmo Marconi (1932)
The Vision People and their machines should be
able to access information and communicate with
each other easily and securely, in any medium
or combination of media-voice, data, image,
video, or multimedia-any time, anywhere, in a
timely, cost-effective way Dr. George H.
Heilmeier IEEE Communication
Mag. October 1992
58
Wireless Overlays
(borrowed from Kerzs talk)
59
Overview of Cellular Systems

• The wireless communication of the future will
  utilize cellular techniques
• Why use cellular technology?
• limited spectrum available
• demand-assigned channel
• allows frequency reuse
• Three basic methods by which cellular carriers
  could make use of the bandwidth that they are
  allowed
• FDMA (Frequency Division Multiple Access)
• TDMA (Time Division Multiple Access)
• CDMA (Code Division Multiple Access)

60
Three Multiple Access Methods
Amplitude
FDMA
Time
1
2
1
2
Frequency
F1
F2
F1
F2
Amplitude
Amplitude
TDMA
CDMA
Time
Time
1
3
1
1
2
2
2
1
3
3
4
4
Frequency
F1
Frequency
F1
F2
F2
F1
F1
61
Cellular Principle

• The cellular technology increases the network
  capacity. It relies on the concept of concurrency
• Concurrency is created by reusing channels in
  different cells i.e., channel/reuse. This is
  allows increase in total capacity of the system
  (i.e., the number of supported users)
• The total coverage area is divided into cells. In
  each cell, only a subset of all the channels is
  available
• All the channels are partitioned into sets, which
  are assigned to cells. The same set if assigned
  to two cells that are geographically distant
  enough, so that the interference between the
  co-channel cells is very small.

62
Cellular Principle (Cont)
63
Cellular Principle (Cont)
64
Cellular Network Structure
System Database
Mobile Terminal
Base Station
MTSO
PSTN
Radio Link
Local Exchange
Switching Control
Network Intelligence
65
Cellular Architecture
66
FDMA AMPS (Advanced Mobile Phone
System)

• AMPS provides the basement of cellular technology
• The total spectrum is divided into channels
  channels are assigned to users for the duration
  of a call
• Cellular phones use a full-duplex channel
• Forward (downlink) channel from BS to MT 869 to
  894 MHz
• FCC (Forward Control Channel) broadcast
  channel, used for subscriber paging and voice
  channel assignment
• FVC (Forward Voice Channel) dedicated channel
  for a single call
• Reverse (Uplink) channel from MT to BS 824 to
  849 MHz
• RCC (Reverse Control Channel) random access
  with sensing provided by FCC
• RVC (Reverse Voice Channel) dedicated channel
  for a single call and paired with the FVC

67
Frequency Allocation ( in Korea)
TRS
AM Radio
FM
TV
Cellular
526.5kHz
851
88MHz
108
470MHz
752
824MHz
849
851
866
Cellular
CT-2
PCS
PCS
869MHz
894
910MHz
1750MHz
1840MHz
914
1780
1870
68
AMPS (Cont.)

• 25 MHz is split into channels that are 30 kHz
  wide
• (24 kHz of which is used exclusively for voice)
• channels 800-900 are not used
• 832 total number of full-duplex channels includes
  21 control channels and 395 voice channels
• 416 channels each between RCC (Radio Common
  Carrier) and WCC (Wireless Common Carrier)

824 - 849 MHz
869 - 894 MHz
...
...
...
...
991
1023
991
1023
2
1
799
2
1
799
Reverse channels
Forward channels
69
AMPS (Cont.)

• 416 channels are divided among a number of cells
  that are designed so that adjoining cells overlap
  slightly (59 or 60 channels for each cell)
• The number of cells among which all of the
  channels are assigned, but none is repeated, is
  called a group
• The configuration would be repeated, reusing the
  frequencies, how?

824
846.5
845
835
825
849
869
891.5
890
880
870
894
991-1023
1-333
334-666
667-716
717-799
991-1023
1-333
334-666
667-716
717-799
A
B
A
B
A
A
B
A
A
B
Reverse channels
Forward channels
70
Reuse Pattern of 7
Maximum number of simultaneous calls total
number of channels (e.g., 416)
3 x 7 x (416 7) 3 x 416
71
Propagation Characteristics

• Two different antennas
• RSA (Rural Service Area) uses omnidirectional
  antennas to cover the maximum amount of area per
  cell
• MSA (Metropolitan Service Area) face with meeting
  the demands imposed by a concentrated customer
  base (120 degrees of a circle)
• Capacity the number of subscribers that may be
  used
• the number of calls placed by the systems
  subscribers
• the way the calls placed are distributed in time
• the average time per call
• the amount of frequency reuse utilized
• the size of cells

72
Cell Splitting

• More capacity vs. more handoff (infrastructure)
• Wireless characteristics
• partial loss, fading, doppler effect

f2
f1
f3
f7
f6
f4
f5
Example Cell Radius 1 mile number of cells
32 (48 ch. / cell) gt 1536 concurrent calls
Cell Radius 0.5 mile number of cells 128
(48 ch. /cell) gt 6144 concurrent calls
73
Cellular Technology has

• Advantages
• more capacity
• less transmission power
• more predictable propagation environment
• more robust system
• Disadvantages
• need more infrastructure (more base-stations)
• need network (to interconnect the base-stations)
• residual interference
• handoffs
• hot sport in user concentration

74
Handoff

• Handoff (U.S. cellular standards) is a procedure
  of changing the MT to BS binding from one BS to
  another BS, according to the MTs move
• cf) handover(CCITT/CCIR), ALT(Automatic Link
  Transfer, ANSI)
• It procedure provides means for improving the
  quality (e.g., RF signal strength) of the
  received signal, while the MT moves or when
  reception conditions change
• The trigger for handoff is the RF signal strength
  falling below some threshold, and when there is
  another BS that can serve the MT with a stronger
  signal
• Soft handoff is a handoff in which the new
  binding is completed before the old binding is
  torn down

75
Handoff (Cont.)

• A cell has three distinct regions with different
  hand-off circumstances
• central region (1) the area in which a need for
  hand-off would be caused by a deep fade
• twofold region (2) two cell overlap
• threefold region (3)
• When handoffs are
• not possible?

76
AMPS Hand-off Routine
MR
Cell A (old)
Cell B (new)
MTSO
Talking
Talking
Degradation of corner level (V-Ch, M-wire)
Carrier level check order
To another cell sites
Carrier level Check order
Carrier level Check with RLR
Carrier level Check with RLR
Level response
Level response
Select the best Cell and an idle Ch.in
the selected cell
Hand-off message
V-Ch
TX-on
SAT-on
Hand-off message
Audio mute
Audio mute
10 Kb/s (v-Ch)
ST on (old V-Ch)
Carrier off change to new V-Ch
ST off
Old Ch on Hosk
TX off
TX-on (Ch skip)
Carrier on at new V-Ch
(V-Ch Ch. Completion E-wire)
Change to new V-Ch
TX off
Talking
Talking
77
Complexity and Cost of Cellular System

• The cost that the consumer sees first
• the user likes the idea of getting something for
  nothing
• the provider will more than recoup its
  expenditure in its service charges
• The cost of the network components is an
  important consideration for the system operator
  the cost of the system must be justified by
  increase in capacity
• if the antennas are designed to handle only a
  sector of a cell, the cost of the switching
  equipment becomes higher
• The quality of service for cellular system is
  usually the percentage of blocked and dropped
  calls

78
TDMA Techniques

• With the proliferation of computers, digital
  communication technologies (i.e., DSP) have
  advanced rapidly
• Allows multiple users to share bandwidth by
  giving a slice of time to each user for
  transmitting and receiving data
• One of the inherent complexities is
  synchronization
• accurate distance and time delay measurements are
  required in order to compute the correct
  transmission time or time advance
• But several advantages of using TDMA
• burst mode transmission results in lower battery
  power consumption
• increased number of concurrent users
• quality of the voice channel

79
NADC (North American Digital Cellular )

• VSELP (Vector Sum Excited Linear Predictive)
  CODEC
• 8kbps
• Dividing channels of the AMPS system into time
  slots
• the same 30 kHz channels of AMPS are used, but
  each channel is divided into three time slots
• Handoff could occur between time slots within the
  same channel, in addition to the handoff between
  channels
• Both the transmitters and receivers become more
  complex, as timing circuits are needed to ensure
  that a transmitter does not infringe upon another
  time slot within the channel
• As a result, adding capacity becomes more
  expensive for both the cellular carrier and for
  the user

80
GSM (Group Special Mobile)

• Aims to unify the EC by offering a single
  standard so that users can use one phone
  throughout Europe
• Designed with the OSI model in mind
• 125 full-duplex channels for eight users on each
  channel
• the channel bandwidth is 200 kHz
• the data throughput is 270.833 kbps per a channel
• in each time slot, 33 kbps are allocated (13 for
  speech coding (RELP), and 20 for overhead
  signaling)
• TDMA employs advanced measurement techniques for
  determining the link quality and the best cell
  for handoff
• more expensive equipment and computers for
  processing

81
CDMA

• CDMA allows multiple users to share the same
  frequency by multiplexing their transmissions in
  the code space
• It was envisioned by Qualcomm as a cellular
  system to replace the current AMPS, using the
  same frequency
• Spectrum is divided into a number of 1.25 MHz
  channels
• for each channel, there are 64 orthogonal codes
• CELP (Code Excited Linear Predictive) for speech
  coding with variable rate, 8.55 kbps max, 3.9
  kbps avrage
• Each channel is potentially shared by a number of
  users that all use a different code to modulate
  data in a spread spectrum transmission

82
CDMA (Cont.)

• CDMA paradigm shift
• multiple users on one frequency
• channel is defined by code
• capacity limit is soft
• Three primary techniques
• vocoder (voice compression / decompression)
• interleaving (variable data rate)
• spectrum spread

83
Comparison of FDMA, TDMA and CDMA

• AMPS is by far the cheapest system to construct,
  build, operate and use, but it is limited by the
  number of users
• CDMA is the most complex and costly, but advances
  in computer technology will make it more
  attractive in the future for the increased
  capacity and service quality
• Intermediated shifting CDMA is inserted into a
  portion of the spectrum and used side-by-side
  with the AMPS
• For 15MHz spectrum allocation

Parameter AMPS GSM
CDMA
Channel BW (MHz) 0.03 0.20
1.25
No. of CHs 500 75
11
Effective CHs 500/7 75/3
11/1
Voice calls/CH 1 7.25
2540
84
PCS (Personal Communications Systems)

• In the narrow sense a mobile telephone service
  that is associated with a person instead of a
  place or a vehicle
• Basic requirements
• users must be allowed to make calls wherever they
  are
• the service must be reliable and of good quality
• it must offer a range of service that the users
  need, such as voice, data, fax, paging and even
  video
• In the operational sense the type of wireless
  communication that implements new digital
  microcellular and provides personal services

85
PCS (Personal Communications Systems)

• The principal idea behind PCS
• each individual user have a unique identification
  number
• using the number, a person can be reached at any
  time and at any place, even if a caller does not
  know the location of the called one
• PCS essentially replicates that of a cellular
  network with one major difference, that is,
  microcellular architecture
• smaller size of cells and large number of base
  stations
• numerous handoffs between cells for moving
  callers
• large investment to setting up a PCS service

86
Wireless Roaming Environment

• Heterogeneous (different technology)
• Unified by Internet Protocol

Satellite
Wireless WAN (GPRS, CDMA 1x, IMT2000)
Internet
Wired or Wireless LAN (Bluetooth, IEEE802.11)
Picocell MAN (Cellular LAN)
87
Wireless Roaming Dimension
88
Content

• Getting start
• distributed computing, what means for MC?
• Internet protocol, and its considerations for MC
• Next generation Internet
• Cellular technology overview
• Internet host mobility
• routing optimization

89
Internet Host Mobility Support

• Five major proposals
• MobileIP (Columbia University)
• Virtual Internet Protocol (Sony)
• Multiple Address Approach (Matsushita)
• IP Option Approach (IBM, CMU)
• IP Mobility Support (IETF)
• Main points of view in this lecture
• addressing conventions
• location details
• routing effectiveness

90
Proposal 1 MobileIP

• Designed and implemented by John Ioannidis
  (Columbia Univ.)
• Aims
• allows mobile hosts to keep their address even if
  it moves
• decouples mobile hosts routing scheme from the
  normal IP routing
• The setup
• mobile Hosts (MHs)
• mobile Support Routers (MSRs)
• campus defined by fully-connected MSRs
• IP protocols defined
• IPIP (IP inside IP)
• MICP (Mobile Internetworking Control Protocol)

91
Model

• System model

92
Addressing (I)

• Based on the embedded network concept
• logical ID embedded address (home address)
• physical locator IP address of the MSR which
  currently serves a MH
• Embedded network consists of its own hosts and
  gateways, and has its own addressing and network
  protocol, but uses parts of another existing
  networks as its infrastructure (it is called as
  local network)
• An embedded address is a two-level construct
  (m,h), where m is the networks identifier and h
  is the hosts identifier
• Virtual network a set of subnets which consists
  of MSRs and their MHs under an admin. control,
  and which therefore share the same m

93
Addressing (II)

• For the MHs controlled by a virtual network, only
  the home addresses are used - the addresses are
  immutable even if the MH moves around
• Each MSR maintains the home addresses of MHs
  under its control and the IP addresses of MSRs
  within the virtual network, and is responsible
  for last-hop delivery to MHs within its service
  area
• How the home address can be mapped into the
  current MSR?

94
Location

• Uses a proxy-ARP between MSRs
• A source host sends a datagram to its current MSR
• If the current MSR does not know which MSR is
  currently responsible for a destination, it
  broadcasts location search queries, using control
  message MICP_WHOHAS (with the destinations home
  address), to all other MSRs of the virtual
  network
• The current MSR of the destination responds a
  control message MICP_IHAVE, including its IP
  address
• The current MSR of the source uses the IP address
  to tunnel the datagram to the destination
• Imagine the location cost, and scalability as
  well!

95
Encapsulation

• IPIP encapsulation

Source Add Sources Current MSR
New IP Header
Dest. Add Dest.s Current MSR
Old IP Header
IP Header
IP Payload
IP Payload
96
Routing

• MHs in the same cell direct routing using ARP
• MHs in different cells tunneling using
  proxy-ARP
• MH to fixed host routed through MSR
• Fixed host (or host outside of the virtual
  network) to MH
• datagrams routed to one of MSR
• the MSR locate the destination MH
• the MSR tunnels the datagrams to the current MSR
• the current MSR delivers them locally, using ARP
• Always pass through an optimal route
• at the expense of heavy network traffic
• also, with taking much time delay

97
Routing Example

• In order to deliver a datagram from MH s to MH k

Data from outside of the virtual network
Virtual Network
MICP_WHOHAS
MICP_IHAVE
subnet a
subnet k
subnet s
MSR p
MH k
MH s
Control Data Data tunneling
98
Popup Operation

• How it will manage in the cases of inter-campus
  mobility?
• MH gets an embedded address from the current
  campus when it newly connect to the current
  campus
• MH notifies the embedded address to an MSR in its
  campus - the MSR is called a designated MSR for
  the MH
• the designated MSR acts as a member of the
  current campus, and it treats the MH as if it
  serves locally
• datagrams for the MH firstly arrive its campus,
  and the receiving MSR tunnels to the designated
  MSR, after identifying the designated MSR
• the designated MSR again tunnels the datagrams to
  the current MSR, after identifying the current
  MSR, using a proxy ARP at the current campus

99
Proposal 2 Virtual Internet Protocol (VIP)

• Designed and implemented by Fumio Teraoka (Sony)
• Main subject in the WIDE project
• Key Ideas
• virtual internet protocol
• propagating cache method
• The setup
• migrating host
• gateways

TCP
UDP
VIP
IP
100
Addressing (I)

• Based on the virtual network concept
• logical ID virtual network address (immutable)
  - usually its own add.
• physical locator physical network address
  (which is a temporary one assigned by the
  subnetwork which a host is currently visiting)
• Virtual networks are logically constructed above
  the physical network by assigning two different
  IP address to each host
• The IP layer then is split into two sublayer
• virtual IP sublayer address mapping between the
  two addresses
• physical IP sublayer conventional IP layer

101
Addressing (II)

• The transport layer specifies the target host by
  its virtual IP address
• A packet sent by a mobile host that is away from
  its home subnetwork carries both addresses
• The Physical IP source and destination addresses
  are conveyed in the conventional IP header,
  whilst the virtual ones are carried either as an
  encapsulated format or as an IP option

102
Datagram Header Format

• When VIP is implemented as an IP option

16
31
0
Vers
Len
Service Type
Total Length
IP Identification
Flags
Fragment Offset
Time to Live
Protocol Num.
Header Checksum
Source IP Address
Destination IP Address
Option length
Option Type
Hold Time
Type
Source VIP Address
Destination VIP Address
Source Address Timestamp
Destination Address Timestamp
Option Type 140
Timestamps acts as a version number
103
Location

• Uses a propagating cache method
• each host and gateway has a cache for address
  resolution
• the cache is called the AMT(Address Mapping
  Table)
• AMT entries are updated/created by two control
  packets, connection/disconnection notification
• AMT entries propagate across the network as data
  communication progresses, i.e. with precisely
  finding out the VIP header
• VIP packet types
• VipData normal data packet
• VipConn connection notification
• VipConnAck ack. of VipConn
• VipDisc disconnection notification
• VipDelAmt AMT entry deletion request
• VipErrObs error notification

104
Connection / Disconnection

• Connection to a subnetwork
• a temporary address is assigned to connecting
  host
• the MH sends a VipConn packet to its home gateway
• intermediate gateway create an AMT entry for the
  MH
• the home gateway broadcasts the VipConn packet in
  the home network and returns a VipConnAck to the
  MH
• Disconnection from a subnetwork
• the MH sends a VipDisc packet to its home gateway
• the home gateway broadcasts a VipDelAmt packet
• if a gateway, which received the VipDelAmt, has
  an AMT entry for the MH, it deletes the
  corresponding entry and broadcasts the packet
• the migrating host releases the IP address

105
Model / Location

• Connection to a subnetwork

Gw-EF
Net-G
Net-E
Net-F
Host-X
Gw-CG
Gw-BF
Host-X
Gw-CD
Gw-BC
Gw-AB
Net-A
Net-B
Net-D
Net-C
Gw-AH
Host-Y
Net-H
Connection Notification Packet
Ack
106
Disconnection

• Disconnection from a subnetwork

Gw-EF
Net-G
Net-E
Net-F
Host-X
Gw-CG
Gw-BF
Host-X
Gw-CD
Gw-BC
Gw-AB
Net-A
Net-B
Net-D
Net-C
Gw-AH
Host-Y
Net-H
Disconnection Notification Packet
AMT Deletion Request Packet
107
Routing (I)

• When a host communicates with a migrating host,
• each host or gateway acts as
• upon reception
• create/update the AMT entry for the source if
  necessary
• before transmission
• if destinations ATM entry exists,
• destinations IP address is resolved
• else assume the IP VIP

108
Routing (II)

• Packet forwarding

Gw-EF
Net-G
Net-E
Net-F
Host-X
Gw-CG
Gw-BF
Host-X
Gw-CD
Gw-BC
Gw-AB
Net-A
Net-B
Net-D
Net-C
Gw-AH
Newly build a cache entry for Host-X
Host-Y
Packet with incorrect PN-address
Net-H
Packet with correct PN-address
Response packet from Host-X
109
Proposal 3 Multiple Addresses Scheme

• Designed and implemented by Hiromi Wada et al.
  (Matsushita)
• Key Ideas
• Packet Forwarding Server (PFS)
• autonomous forwarding mode
• Addressing
• logical ID home IP address (immutable)
• physical locator temporary IP address (which
  is assigned by the subnetwork which a host is
  currently visiting)
• Location
• each subnetwork has at least one special router,
  PFS
• the PFS is responsible for tracking the temporary
  IP address
• the new temporary address for a mobile host
  should be notified from the host itself to its
  home PFS and the previous PFSs which have been
  just left by the host

110
Model / Location

• MH ks move from subnet m to subnet k, then to
  subnet a

Internet
To Previous PFS
To Home PFS
subnet s
subnet a
subnet m
subnet k
PFS a
PFS k
PFS m
PFS s
SH s
SH m
MH k
host moving
host moving
Location notification
111
Datagram Header Format)

• IPTP (Internet Packet Transmission Protocol)
  encapsulation

16
31
0
Vers
Len
Service Type
Total Length
IP Identification
Flags
Fragment Offset
Time to Live
Protocol Num.
Header Checksum
Source IP Address
Destination IP Address
Type
Sequence
Aim
Status
(not used)
Counter
Autonomous
Home Address of MH
Temporary Address of MH
Address of PFS
Type
0 Packet transmission message
1 MH Locat