Mobile Computing and Databases A Survey Daniel Barbar

1
Mobile Computing and Databases A SurveyDaniel
Barbará

• Pallavi Phene
• phene_at_cs.uh.edu
• November 12th, 2002

2

• People and their machine should be able to
• access information and communicate with
• each other easily and securely, in any
• medium or combination of mediavoice,
• data, image, video, or multimediaany time,
• anywhere, in a timely, cost-effective way
• Dr. George H. Heilmeier
• IEEE Communication Mag.
• October 1992

3
MANETsMobile Ad-hoc NETworks

• Networks formed by Wireless, Mobile hosts without
  (necessarily) using a pre-existing infrastructure.

4
Problems

• a) Routes between nodes may potentially contain
  multiple hops

5
Problems

• b) Hidden Terminal Problem

C
A
B
6
Problems

• c) Near Far Problem
• Near-by terminals over power signals from
  far-away terminals

BS
A
B
C
7
Problems

• d) Hand-offs

Base Stations (BS)
Mobile Host (MH)
8
Effect of Mobility

• Physical Layer
• Channel varies with user location and time
• Data Link Layer
• Reliable communication interrupted by bursts
• Network Layer
• Re-routing due to movement of hosts
• Presentation Layer
• Source coding for efficiency
• Application Layer
• Location Dependent Applications

9
Mobile Computing

• In mobile networking, computing activities are
  not disrupted when the user changes the
  computer's point of attachment to the Internet.
  Instead, all the needed reconnection occurs
  automatically and non-interactively.
• Charles E. Perkins Sun Microsystems

10
Outline

• Characteristics of Wireless Networks
• Mobile Computing and Databases
• Data Dissemination
• Data Consistency
• Location Dependent Querying
• Interfaces for Browsing
• Challenges

11
A Mobile Environment
12
Outline

• Characteristics of Wireless Networks
• Mobile Computing and Databases
• Data Dissemination
• Data Consistency
• Location Dependent Querying
• Interfaces for Browsing
• Challenges

13
Characteristics of Wireless Networks

• Asymmetry in Communications
• Frequent Disconnections
• Power Limitations
• Screen Size

14
Outline

• Characteristics of Wireless Networks
• Mobile Computing and Databases
• Data Dissemination
• Data Consistency
• Location Dependent Querying
• Interfaces for Browsing
• Challenges

15
Mobile Computing and Databases

• Asymmetry Power Limitations
• Data Dissemination
• Asymmetry Frequent Disconnections
• Transaction Mgmt and Data Consistency
• Frequent Disconnections
• Location Dependent Querying
• Screen Size
• Interface for Browsing

16
Outline

• Characteristics of Wireless Networks
• Mobile Computing and Databases
• Data Dissemination
• Data Consistency
• Location Dependent Querying
• Interfaces for Browsing
• Challenges

17
Data Dissemination

• It is a push-based model for broadcasting data to
  the client rather than waiting for the client to
  request the specific data.

18
Data Dissemination

• Advantage
• no interruptions by the client requests
• send additional related data
• Disadvantage
• figuring out the relevant additional data
• Depends
• On ability of the server to predict the clients
  needs

19
Types of push-architectures

• Periodic Push
• delivery is performed on a regular, repeating
  schedule
• client can disconnect and still not lose data
• Aperiodic Push
• has no such schedule
• assumes that clients are either always listening
  and able to respond to what is being sent
• makes more effective use of the downstream
  communication channel

20
Methods of Data Dissemination

• Broadcast Disks
• Interleaved Push and Pull (IPP)
• Invalidation Reports (IR)

21
Broadcast Disks

• It is a periodic dissemination architecture
• Multiple disks of different sizes are
  superimposed on the broadcast medium (Server
  Broadcast Programs)
• Exploits the client storage resources for caching
  data (Client Cache Mgmt)

22
Server Broadcast Programs

• Server takes the union of required items (pages)
  and broadcasts the resulting set cyclically.
• Number of disks (num_disks) determine the number
  of different frequencies with which pages will be
  broadcast.
• For each disk, the number of pages (num_pages(i))
  and the relative frequency of broadcast
  (rel_freq(i)) are specified.

23
An Example

• num_disks 3
• Disk 1
• num_pages(1) 1
• rel_freq(1) 4
• Disk 2
• num_pages(2) 2
• rel_freq(2) 2
• Disk 3
• num_pages(3) 8
• rel_freq(3) 1

24
Client Cache Management

• Clients cache the pages for which the local
  probability of access is higher than the
  frequency of broadcast.

25
Interleaved Push and Pull (IPP)

• Integration of a pull-based and a push-based
  Broadcast Disk approach.
• Communication Bandwidth shared between 2 Channels
• Front-Channel
• used by servers for push-based operations i.e.
  for maintaining Broadcast disks and providing
  responses to client pull operations
• Back-Channel
• used by clients for pull-based operations

26
Improving Scalability of IPP

• Adjusting pull bandwidth
• Providing a Pull threshold
• Increase available pull bandwidth by chopping off
  the slowest part of the broadcast schedule

27
Invalidation Report (IR)

• The server uses IRs to notify clients about
  changes in the items being cached by the clients
• IRs contain only the changes in the data values
  Static/unchanged data is not re-sent
• IRs Can cause false-negatives
• Quasicopies and adaptive IRs may be used

28

• IRs can be used to check the validity if the
  items/groups in the clients cache using GCoRe
  (Group with Cold Update-Set Retention)
• Can be combined with clients behavior to
  broadcast IRs only with a users mobility area

29
Broadcast Disks in Real-Time Environments

• Real-time environments are time-critical
• Broadcast disks do not accommodate transmission
  failures
• Require a technique to add reliability to the
  Broadcast Disks technique
• Organizations suitable for RT environments
• Flat
• Rate Monotonic
• Slotted Rate Monotonic

30
Organization of Broadcast Disks to Ensure Fault
Tolerance

• Achieved using AIDA (Adaptive Information
  Dispersal Algorithm), which is an elaboration of
  IDA (Information Dispersal Algorithm)
• AIDA uses minimum controlled redundancy to
  guarantee timeliness and fault tolerance up to
  any degree of confidence

31
IDA

• Divides a file into N independent pieces, such
  that combining any m pieces is sufficient to
  retrieve the file

Dispersal and Reconstruction of Information using
IDA
32
AIDA

• In AIDA a bandwidth allocation operation is
  inserted after the dispersal operation but prior
  to transmission
• This allows the system to scale the amount of
  redundancy used in the transmission
• Number of pieces transmitted n is allowed to vary
  from m (no redundancy) to N (maximum redundancy)
• Using bandwidth allocation, the redundancy of
  unimportant items can be reduced and that of
  critical items increased

33
AIDA

• Dispersal and Reconstruction of Information using
  AIDA

34
An Example of a Flat Broadcast Program
35
Another Consideration Directories

• Indexing on Air
• transmitting index of what data items are being
  broadcast
• Advantage
• allows clients to be inactive some of the time
  and hence save power
• Distributed Indexing
• sending only part of the Index
• Methods of Distributed Indexing
• Temporal Addresses
• Multicast Addresses

36
Outline

• Characteristics of Wireless Networks
• Mobile Computing and Databases
• Data Dissemination
• Data Consistency
• Location Dependent Querying
• Interfaces for Browsing
• Challenges

37
Data Consistency

• Consistency requires that data bound by a
  transaction be semantically preserved.

38
Session Guarantees

• Read Your Writes
• Any read operation in the session must reflect
  the values established by previous writes in that
  session
• Monotonic Reads
• Successive reads reflect a non-decreasing set or
  writes
• Writes Follow Reads
• Writes are propagated after the reads on which
  they depend
• Monotonic Writes
• Writes are propagated after writes that logically
  precede them

39
Maintaining Data Consistency

• Certification Reports
• Isolation Only Transaction (IOT)
• Data Replication

40
Certification Reports

• Are used to support transaction management in
  mobile environments
• CRs contain lists of items that are in the read
  and write set of active transactions
• CRs are used by the clients to verify if the
  transactions being run by them need to be aborted
• The server completes the verification only when
  the client cannot detect any conflict
• If the transaction can commit, then the server
  will put the values in the database and notify
  the client

41
Isolation Only Transactions(IOT)

• Guarantees only Consistency of Databases
• When the transaction completes, it enters either
  a committed or pending state.
• If in committed state, then the results are sent
  to the servers to be committed
• If in pending state, the transactions is in wait
  state and is validated later

42
Data Replication

• Required so that processing continues smoothly in
  case of disconnections
• Technique requires referee nodes that update and
  store the core (updateable data items) set
  descriptions
• Other issues
• maintaining replication of directories (Primary
  by Row)
• replication in weakly connected systems (Lazy
  Release Consistency)

43
Outline

• Characteristics of Wireless Networks
• Mobile Computing and Databases
• Data Dissemination
• Data Consistency
• Location Dependent Querying
• Interfaces for Browsing
• Challenges

44
Location Dependent Querying

• Ad-hoc queries will be addressed to large
  databases
• regarding the local area
• e.g. people, places, routes, services etc
• location transparent
• e.g. recent sales figure of a particular
  product, who stores a particular product on his
  PDA etc.

45

• Problem
• Minimizing communication cost to retrieve the
  query result
• Solutions
• integrate GPS into IP to enable creation of
  location independent services
• e.g. Genesis, Advanced Traveler Information
  Systems (ATIS), Mobisaic (using Dynamic URLs and
  Active Documents)

46
Outline

• Characteristics of Wireless Networks
• Mobile Computing and Databases
• Data Dissemination
• Data Consistency
• Location Dependent Querying
• Interfaces for Browsing
• Challenges

47
Interfaces for Browsing

• Query by Icons (QBI)
• query processing facility supporting exploration
  and querying of databases from a mobile computer
  based on the manipulation of icons
• addresses the screen size, memory and battery
  power, communication bandwidth limitations
• Features
• Iconic visual language
• Semantic Data Model
• Meta-query tools

48
Architecture of QBI
49
QBI Interface
50
Query Example
51
Outline

• Characteristics of Wireless Networks
• Mobile Computing and Databases
• Data Dissemination
• Data Consistency
• Location Dependent Querying
• Interfaces for Browsing
• Challenges

52
Challenges

• Prototyping
• Bandwidth Utilization
• Transactional Properties
• Optimization of Location Dependent Query
  Processing
• Data Visualization

53
Thank You!