mds_khazaei

Mobile Database SystemsMobile Database Systems

Hamzeh KhazaeiHamzeh KhazaeiMath & Computer Sc. Dep Math & Computer Sc. Dep

Amirkabir University of Amirkabir University of Technology, (Tehran Technology, (Tehran

Polytechnic)Polytechnic)[email protected]

Agenda

Mobile Database Systems, Intro Wireless Network Communication Location & Handoff Management Fundamentals of Database Technology Concurrency Control Mechanisms Data Processing & Mobility Transaction Management in MDS Mobile Database Recovery Wireless Information Broadcast


Outline

Fully Connected Information Space Personal Communication System (PCS) Mobile Database Systems (MDS) Transaction Management Data Caching Query Processing Data Classification Conclusion


Fully connected information spaceFully connected information space


Fully connected information space

Each node of the information space has some communication capability.

Some node can process information.

Some node can communicate through voice channel.

Some node can do both


Fully connected information space

Can be created and maintained by integrating

legacy database systems, and wired and wireless

systems (PCS, Cellular system, and GSM)


A system with the following structural and functional properties

Distributed system with mobile connectivity

Full database system capability

Complete spatial mobility

Built on PCS/GSM platform

Wireless and wired communication capability

What is a Mobile Database System (MDS)?


What is a mobile connectivity?

A mode in which a client or a server can establish

communication with each other whenever needed.

Intermittent connectivity is a special case of mobile

connectivity.


A node in which only the client can establish

communication whenever needed with the server but the

server cannot do so.

What is intermittent connectivity?

Types of Mobility

Terminal Mobility

Types of Mobility

Personal Mobility

Terminal Mobility

It allow a Mobile unit (laptop, cell phone, PDA, etc) to access desired services from any location while in motion or stationary, irrespective of who is carrying the unit.

For example a cell phone can use of network service with owner or someone else.

In terminal mobility, it is the responsibility of wireless network to identify communication device.

Personal Mobility

A user dose not have to carry any communication equipment with him.

User can use any communication device for establishing communication with the other party.

This facility requires an identification scheme to verify the person wishing to communicate.

Each person has uniquely identified and can use any mobile device for communication. (example is the internet).

Wireless Network Communication

This part deal with architecture of cellular communication and identifies hardware & software component to build mobile database systems.

All wireless system use electromagnetic waves to propagate information from one place to another. (radio frequencies).

Use different frequencies for different purposes.

Cellular Communication

Geographical area divided to some slots that use RF to setup communication in each slots, this slots referred to as a cell.

The entire communication infrastructure is known as cellular communication.

The best coverage pattern is hexagons shapes, that no have uncovered space.

A cell must have a wireless component for managing the communication.

The transceiver is installed in the best place of cell.


In a cellular architecture a number of wireless and wired components are required to establish the desired point-to-point, point-to-multipoint communication.

One such component is the transceiver, which referred as Base Station (BS).

A BS functions under the supervision of telecommunication switch called Mobile Switching Center (MSC) and connected to it through wired line.

Structure of a Channel

In mobile discipline two frequencies are required to establish communication; one from mobile unit (MU) to Base Station (BS) (uplink channel) and inverse (downlink channel).

Downlink and uplink Channel

PCS & GSM

Personal Communication SystemEvolved in North America Support Digital and analogue

Global System for Mobile CommunicationEvolved in Europe Purely Digital

Personal Communication System (PCS)Personal Communication System (PCS)

Architecture Wireless communication Bandwidth limitations Frequency reuse

Part 1


A system where wired and wireless networks are integrated for

establishing communication.


Wireless Components

BS

MSC (MTSO)

MSMS Wireless component

Base Station (BS): A switch, which serves as

communication link between MU and the entire

network

Mobile Units (MU): Also called Mobile Systems (MS) or

Mobile Hosts (MH). A mobile component, which

communicates with BS through a limited number of

wireless channels.


Wireless Components

Home Location Register (HLR): It’s a large database

which stores necessary information such as

geographical location of subscriber, call processing,

billing, service subscription, service restriction, etc. It

can be distributed database.

Authentication Center (AC): The AC is the processor

system, witch authenticate subscribers, AC needs to

access user information for authentication process so

its co-located with HLR, then the AC and HLR store in

MSC. The authentication steps as follows :


Wireless Components

1- AC sends a random number to the mobile unit from

where the call originates.

2- Authentication algorithm stored in SIM (Subscriber

Identity Module) manipulates this random number using

a Subscriber Authentication Key, which is also stored in

SIM.

3- The result of this manipulation is sent to AC along with

an encryption key.

4- Concurrent with the authentication computation at the

mobile unit, AC performs identical computation using the

random number and information stored in HLR.


Wireless Components

5- AC compares the result of its own computation and the

result received from the mobile unit. In the case of

successful comparison, it permits the subscriber to access

the network and stores and sends the encryption key to BS

to enable ciphering to take place.

Equipment Identify Register (EIR): it is a database which

stores information for the identification of mobile units. It

maintains a database of Electronic Serial Number which is

unique to mobile unit which prevent its theft and malicious

use.


Wireless Components

Public Switched Telephone Network (PSTN) : This

component refers to the regular wired line

telecommunication network which is commonly

accessed by landline calls.

Integrated Service Digital Network (ISDN): It is a wired

line network which provides enhanced digital services

to subscribers.

Short Message Entity (SME): This is a part of the

System Message Service (SMS), which looks after text

messages.


Wireless Components

Message Center (MC) : This unit stores and forwards

short messages to mobile destination. If the

destination is unavailable for any reason , it stores the

message for later dispatch.


Limited channels must be utilized efficiently. It is done

so by

Frequency reuse

The same radio frequency is used for

communication by more than one cell

sessions.

Mobile cells

To achieve frequency reuse, the entire

wireless coverage area is divided into

cells.


Mobile cellsMetropolitan area Metropolitan area

Coverage area in one cell Coverage area in three cells

BS

BSBSBase Station

Large cells.Low density

Small cells.High density

Smaller cells.Higher density


Mobile cells

The entire coverage area is a group of a number of

cells. The size of cell depends upon the power of

the base stations.

PSTNMSC


Frequency reuse

61

7

54

3

2

61

7

54

3

2

61

7

54

3

2

D A

AA

AA

AA

NR

D3

D = distance between cells using the same frequencyR = cell radiusN = reuse pattern (the cluster size, which is 7).

Thus, for a 7-cell group with cell radius R = 3 miles, the frequency reuse distance D is 13.74 miles.


Problems with cellular structure

How to locate of a mobile unit in the entire coverage area?

Solution: Location management

How to maintain continuous communication between two parties in the presence of mobility?

Solution: Handoff

How to maintain continuous communication between two different service provider?

Solution: Roaming


HandoffA process, which allows users to remain in touch, even

while breaking the connection with one BS and

establishing connection with another BS.

Old BS New BS

MSC

Old BS New BS

MSC

MSC

Old BS New BS New BSOld BS

MSC


Handoff

To keep the conversation going, the Handoff

procedure should be completed while the MS (the

bus) is in the overlap region.

G

Old BS New BS

Cell overlap region


Handoff issues

Handoff detection

Channel assignment

Radio link transfer


Mobile-Controlled handoff (MCHO)

Network-Controlled handoff (NCHO)

Mobile-Assisted handoff (MAHO)

Handoff detection strategies


Mobile-Controlled Handoff (MCHO)

In this strategy, the MS continuously monitors the

radio signal strength and quality of the surrounding

BSs. When predefined criteria are met, then the MS

checks for the best candidate BS for an available

traffic channel and requests the handoff to occur.

MCHO is used in DECT and PACS.


Network-Controlled Handoff (NCHO)

In this strategy, the surrounding BSs, the MSC or

both monitor the radio signal. When the signal’s

strength and quality deteriorate below a predefined

threshold, the network arranges for a handoff to

another channel. NCHO is used in CT-2 Plus and

AMPS.


Mobile-Assisted Handoff (MAHO)

It is a variant of NCHO strategy. In this strategy,

the network directs the MS to measure the signal

from the surrounding BSs and to report those

measurements back to the network. The network

then uses these measurements to determine

where a handoff is required and to which channel.

MACHO is used in GSM and IS-95 CDMA.


Handoff types with reference to the network

Intra-system handoff or Inter-BS handoff

The new and the old BSs are connected to

the same MSC.

Old BS New BS

MSC


Handoff types with reference to the network

Intersystem handoff or Inter-MSC handoff

The new and the old BSs are connected to

different MSCs.

Old BS New BS

MSCMSC


Handoff types with reference to link transfer

Hard handoff

The MU connects with only one BS at a time,

and there is usually some interruption in the

conversation during the link transition.

Soft handoff

The two BSs are briefly simultaneously

connected to the MU while crossing the cell

boundary. As soon as the mobile's link with

the new BS is acceptable, the initial BS

disengages from the MU.



Hard handoff

1. MU temporarily suspends the voice conversation

by sending a link suspend message to the old BS.

2. MU sends a handoff request message through an

idle time slot of the new BS to the network.

3. The new BS sends a handoff ack message and

marks the slot busy.

4. The MU returns the old assigned channel by

sending a link resume message to the old BS.



Hard handoff

5. MU continues voice communication while the

network prepares for the handoff.

6. Upon receipt of a handoff request message, the

new BS sends a handoff ack message and

reconfigures itself to effect the handoff.

7. The MSC inserts a bridge into the conversation

path and bridges the new BS.

8. Finally, the network informs the MU to execute the

handoff via both the new and old BSs by sending

the handoff execute message.



Hard handoff

9. MU releases the old channel by sending an

access release message to the old BS.

10. Once the MU has made the transfer to the new

BS, it sends the network a handoff complete

message through the new channel, and resumes

the voice communication. The network removes

the bridge from the path and frees up the

resources associated with the old channel.



Soft handoff

1. MU sends a pilot strength measurement message

to the old BS, indicating the new BS to be added.

2. The old BS sends a handoff request message to

the MSC. If the MSC accepts the handoff request,

it sends a handoff request message to the new

BS.

3. The BS sends a null traffic message to the MU to

prepare the establishment of the communication

link.



Soft handoff

4. The new BS sends a join request message to the

MSC. The MSC bridges the connection for the

two BSs, so that the handoff can be processed

without breaking the connection.

5. The new BS sends a handoff ack message to the

old BS via the MSC. The old BS instructs the MU

to add a link to the new BS by exchanging the

handoff command and handoff complete

messages.



Soft handoff

6. The old BS and the MSC conclude this procedure

by exchanging the required handoff information.

The quality of the new link is guaranteed by the

exchange of the pilot measurement request and

the pilot strength measurement message pair

between the MU and the new BS.


Roaming

Roaming is a facility, which allows a subscriber to

enjoy uninterrupted communication from anywhere in

the entire coverage space.

A mobile network coverage space may be managed

by a number of different service providers. They

must cooperate with each other to provide roaming

facility.

Roaming can be provided only if some administrative

and technical constraints are met.


Roaming

Administrative constraints

Billing.

Subscription agreement.

Call transfer charges.

User profile and database sharing.

Any other policy constraints.


Roaming

Technical constraints

Bandwidth mismatch. For example, European

900MHz band may not be available in other

parts of the world. This may preclude some

mobile equipment for roaming.

Service providers must be able to

communicate with each other. Needs some

standard.

Mobile station constraints.


RoamingTechnical constraints

Integration of a new service provider into the

network. A roaming subscriber must be able

to detect this new provider.

Quick MU response to a service provider’s

availability.


Location Management

Two-Tier Scheme

HLR: Home Location Register

A HLR stores user profile and the

geographical location.

VLR: Visitor Location Register

A VLR stores user profile and the current

location who is a visitor to a different cell that

its home cell.


Location Management

Two-Tier Scheme steps. MU1 wants to talk to MU2.

MU1

MU2

Cell 1 Cell 2


Location Management

Two-Tier Scheme steps. MU1 wants to talk to MU2.

1. VLR of cell 2 is searched for MU2’s profile.

2. If it is not found, then HLR is searched.

3. Once the location of MU2 is found, then the

information is sent to the base station of cell 1.

4. Cell 1 establishes the communication.


Location Management

Two-Tier Scheme steps location update

1. MU2 moves from cell 1 to cell 2.

2. MU2’s location is changed so new location

must be recorded.

3. HLR is updated with the new location address.

4. MU2’s entry is deleted from the VLR of cell 1

and new entry is made in cell 2’s VLR.


Location ManagementTwo-Tier Scheme steps location search

Location Search

1- The Caller dials a number, to find the called number (destination), the caller unit sends a location query to its base station source base station.

2- the source base station sends the query to the S-LS (source location server) for location discovery.

3- S-LS first looks up the VLR to find the location. If the called number is a visitor to the source base station, then the location is known and connection is set up.

4- If VLR search fails, then the location query is sent to HLR.

5- HLR finds the location of D-LS (Destination Location server)

Location Search

6- The search goes to D-LS7- D-LS finds the address of D-BS (destination

base station)8- Address of D-BS send to the HLR9- HLR sends the address of D-BS to S-LS

(source location server)10- The address of D-BS is send to the source

base station, which sets up the communication session.


Location ManagementTwo-Tier Scheme steps location update

Location Update

1- The mobile unit moves to a new registration area which is serviced by a new location server (New LS). The mobile unit informs the new base station about it’s arrival.

2- The new base station sends the update query to new LS.

3- The new LS searches the address of the HLR in its local database.

4- The new location of the mobile unit is sent to HLR.5- The old location of the mobile unit is replaced by the

new location.

Location Update

6- The HLR sends user profile and other information to new LS.

7-The new LS stores the information it received from HLR.

8- The new LS informs the new base station that location update has been completed.

9- the HLR also send a message about this location update to the old LS. The old LS deletes the old location information of the mobile unit stored in its database.

10- the old LS sends a confirmation message to the HLR.

End of part one

Fundamentals of Database Processing

The types of DBMSCentralized

There is one node processing (server) and a number of intelligent terminals (client) connected to the server with wired network

DistributedA number of centralized database systems are

connected together with a communication network.

Fundamentals of Database Processing

The servers can be configured dependent or independent.

Homogeneous or heterogeneous.

Centralized DBMS

Distributed DBMS

Distributed DBMS

Based on : Node autonomy Data processing Distribution mode

There are two commonly known categories: Federated database system Multidatabase system

Federated architecture

Subset of servers are partially autonomous in managing their activities

They are member of federation and willing to cooperate and participate with other members (servers) under agreed protocols.

This federation can be homogeneous or heterogeneous.

example is a big company that have many department that each have their own server, and participate themselves to maintain the consist view of the entire company.

Multidatabase architecture

All servers have full autonomyIt’s up to server to cooperate or not with

any other server.

Database partition and distributionIn distributed database the database is distributed in

three ways: PartitionedThe entire database is distinctly divided into a number

of partitions.Some important criteria for partitioning:

Support highest database locality Minimize the cost of communication Minimize the cost of maintaining global consistency Minimize of recovery

Partial Replication

Database is partitioned and a subset of partitions is replicated at more than one servers.

Improve database locality and reliabilityThe recovery is easier than partitionedMore time consuming to maintain the

global consistency

Full Replication

The entire database is replicated at all servers

Maximum locality and minimize data communication cost

Highest reliability and availability Maximum cost of maintaining global

consistency. This schema not used reality, because of

higher consistency and storage cost.

Transaction

Transaction with it’s property preserve consistency of database.

Constraint can be implement in two ways:Programmed in the application codeIncluded in transaction structure

Flat Transaction

Is the simplest type of mechanism which support one level operation

During of their execution, they do not trigger other dependent transaction

It has for essential property:Atomicity (idempotent property of rollback)Consistency Isolationdurability

Advanced transaction model

For achieve :Increase the degree of parallelism in the

execution of long running transactionRedefine consistency to manage

statistical database transactionHandle failure cases in confined wayRedefine conflict scenarioAdvanced transaction necessary.

Nested transaction model


Sub transactions at the leaf are flat transaction which actually perform the necessary data processing

Parent does not manipulate any data, but invokes sub transactions and controls the flow of execution.

A sub transaction can be commit or rollback independent to other sub transactions or the parent of the tree.

A sub transaction has only A,C and I properties, because it’s durability depends only on its parent.

If the parent rolls back, then all its sub transaction must be rolled back.


Open nested transaction modelThe parent does not enforce restriction of

the execution, rollback, and commitment of its sub transactions.

The parent only invoke sub transaction, and the sub transactions executed independently to each other and also to the parent.

Closed nested transaction model

SAGA transaction model

SAGA based on compensation transaction and long running transaction.

A transaction can be divided into number of fine granularity transaction and chained together, then each fine transaction can execute independently.

It’s also possible to remedy (compensation) the malfunctioning of a single fine transaction without effecting others.

Each sub transaction correspond with the compensation transaction.

SAGA with ACID

SAGA allows other SAGAs to see the partial results of a transaction, as a result SAGA can not guarantee complete atomicity

Consistency and Isolation because of above property is not guaranteed

Durability is preserved because SAGA guarantees that the main transaction commits only when al its subtransaction are committed.

Cooperative transaction model


Transaction Group (TG):A TG is a parent for a number of cooperative

transaction, and supervises the activities of cooperative members.

It interact with other TGs with external protocols.Every TG its own version of a data item which is

not accessible to other TG.All cooperating member can access to this versionTG does not take part in data processingTG plays the role of server for its cooperating

transaction member.


Cooperative Transaction members:Operate under internal protocolThey are like client that issue reads and writes on

data items which performed by their TGTG can either process operation from client

process, refuse and queue for later processingSerialization of member is monitored by TGFlexibility in achieving serialization and purpose-

specific correctness criteria are of the contributions of cooperative transaction model.

Flex Transaction Model

This model is based on flat transaction Allow users to define their data processing

preferences The objective of this model is to minimize

ACID constraints to allow more flexibility in processing concurrent transactions.

A flex transaction T can be represent as 5-tuple (B,S,F,π,f). Each members describe as follows:

Flex Transaction Model

Concurrency Control Mechanisms

Concurrency Control Mechanisms

The available CCMs can be categorized into:Locking (2 phase locking)

Locking and un locking are atomic and mutually exclusive for transaction.

Non locking

Ways of locking data items

Simultaneous locking and simultaneous unlocking.

Incremental locking and simultaneous unlocking.

It can cause of deadlock

simultaneous locking and Incremental unlocking.

It suffer with cascading, which is serious time-consuming activity

Ways of locking data items

The objective is to minimize the transaction waiting. It suffer from deadlock and cascading, so it not use in

reality.

Non locking based schemes

TimestampIn this method the execution order is

defined before they begin their executionThe order of execution is established by

associating timestamp (integer) to every transaction

Simple timestamp schemeBasic timestamping scheme

Optimistic concurrency control

It tries to reduce locking overhead by delaying lock operation until conflicting of transactions are ready to commit.

It uses three phases: read, validate and write. Read: transaction read the desired data completely

unrestricted and save it in local cache Validate: in this phase it is determined that transaction will

not generate an inconsistent data and the result is correct Write: if the validation success, the transaction writes the

modified data to database, to be available for other transaction.

This approach replace lock overhead with rollback overhead in compare with 2PL.

Data processing and Mobility

Mobile Database Systems (MDS)Mobile Database Systems (MDS)

MDS Data Management Issues

How MDS looks at the database data?

Data classification

Location Dependent Data (LDD)

Location Independent Data (LID)




The class of data whose value is functionally

dependent on location. Thus, the value of the

location determines the correct value of the data.

Location Data value

Examples: City tax, City area, etc.



The class of data whose value is functionally

independent of location. Thus, the value of the

location does not determine the value of the

data.

Example: Person name, account number, etc.

The person name remains the same irrespective

of place the person is residing at the time of

enquiry.

Location Independent Data (LID)




Example: Hotel Taj has many branches in India.

However, the room rent of this hotel will depend

upon the place it is located. Any change in the

room rate of one branch would not affect any

other branch.

Schema: It remains the same only multiple

correct values exists in the database.




LDD must be processed under the location

constraints. Thus, the tax data of Pune can be

processed correctly only under Pune’s finance

rule.

Needs location binding or location mapping

function.




Location binding or location mapping can be

achieved through database schema or through

a location mapping table.



Location Dependent Data (LDD) Distribution

MDS could be a federated or a multidatabase

system. The database distribution (replication,

partition, etc.) must take into consideration

LDD.

One approach is to represent a city in terms of a

number of mobile cells, which is referred to as

“Data region”. Thus, Pune can be represented

in terms of N cells and the LDD of Pune can be

replicated at these individual cells.



In a data region the entire LDD of that location

can be represented in a hierarchical fashion.

County 1 data County 2 data County n data

City data

Subdivision 1 data Subdivision data Subdivision m data

Concept Hierarchy in LDD


MDS Query processing

Query types

Location dependent query

Location aware query

Location independent query




A query whose result depends on the geographical

location of the origin of the query.

Example

What is the distance of LA railway

station from here?

The result of this query is correct only for “here”.




Situation: Person traveling in the car desires to

know his progress and continuously asks the

same question. However, every time the answer

is different but correct.

Requirements: Continuous monitoring of the

longitude and latitude of the origin of the query.

GPS can do this.

Effect of Mobility on ACID

Effect of Mobility on Atomicity:It is difficult to enforcement atomicity in

MDSTransaction logs are use for atomicity.In conventional system that logs find on

server easily.In MDS the logs not easy to access

because of disconnect mode of MU.

Effect of Mobility on Consistency

In conventional system there is only one correct value for each data object.

Mutually consistent state if all copies have the exact same value.

In mobile database the presence of location-dependent data defines two types of consistency:Spatial consistencyTemporal consistency

Effect of Mobility on Isolation and durability

In MDS ensure the isolation at the fragment level is enough.

Regional and global durability should be considered.

Effect of Mobility on commit

Transaction commitment not effected by mobility.

Because of location-dependent data, a location commit is defined.

A location commit binds a transaction commit to a region.

Example:Reserve 5 seats in a vegetarian restaurant

located 1 mile from here

Transaction Management in Mobile Database Systems

Transaction management in mobile database systems

PCS and GSM used for establish mobile database

Conventional ACID transaction model was unable to satisfactory manage mobile database because of following reason:Hands off, which unpredictableDoze mode of mobile unitDisconnected modeForce disconnectionLack of memory and channel resourceLocation dependent data

Mobile database systems

MDS provides full database and mobile communication functionalities. It allows mobile user to initiate transaction from any ware and at any time and guarantees their consistency preserving execution. In the any kind of failure (transaction, system and media), MDS guarantees database recovery

An MDS is a distributed multidatabse client/server system based on PCS & GSM

Architecture of MDS


Fixed Hosted (FH) are general purpose computers that interconnected through a high-speed wired network.

FHs are not fitted with transceiver, so they do not communicate with MU

One or more BSs are connected with a Base Station Controller (BSC) that coordinate the BSs to communicate BSs with FHs and DBS

To incorporate full database functionality, its necessary to incorporate DataBase Server (DBS) to PCS or GSM

DBSs are connected to the mobile system through wired line as separate nodes

Each DBS can be reached by any BS or FH A DBS communicates with a MU only through BSs


Three mode of mobile unit:Powered mode (not actively listening to the BS) Idle mode (doze mode – not communicate but

continuously listening to the BS)Active mode (communicating with other party,

processing data) The MDS has multiple DBSs, and a database

can be distributed by partition or by partial or fully replication.

Special replication must follow the constraints of location dependent and location free data

Different replication types

Transaction execution in MDS

Implements parallel processing by fragmenting a transaction into a number of subtransactions which are then executed at multiple nodes

The entire execution require a software modules called a coordinator

The coordinator must have Direct and continuous communication with other nodes Theoretically unlimited and continuous power and large

storage space High reliability and availability

So Between MU, DBS, FH, MSC and BS which one is suitable for coordinator?

Transaction Processing in MDS

A transaction in MDS can be initiated from a DBS or from MU or from both

It can be processed entirely at the MU or in the DBS or a combination of the two

If process entirely at one node, so coordinator have minor role in execution and commitment of the transaction

It is important to note that no other MUs should be involved in the execution of transaction

If an MU initiate transaction, at first check that if the transaction can be entirely executed locally (at MU), in this case the MU execute and commit transaction and send the update for install in server and finally send the result to the user


If it cannot be entirely executed at MU there are two options:Transfer required data items from DBSs to

MUMU get the desire data from BSSend the results to user and update to install in

serverNot require of coordinator servicesSignificant amount of communication overhead


Distribute transaction to a set of nodesThe MU divides the transaction into a number of

subtransactionsKeeps the subtransactions it can executeWith help of coordinator distribute the rest of

subtransactoins to a subset of DBSsThe current BS of the MU becomes the

coordinator of transaction and manage the execution leading either to a commit or abort

This schema generates less communication overhead and database update cost

Mobility and transaction execution

The mobility of MU make the job of coordinator difficult, following processing scenarios: Mobile Unit dose not move MU moves and the transaction entirely execute on MU, this

is quite common in mobile computing Distributed processing and MU moves, transaction

originates at a MU and is fragmented to MU and DBSs The service of coordinator must available

continuously during the execution of transaction This link may be broken when the MU crosses the

current cell boundary and enters a different cell


The link can be maintain in two ways:Static method

Coordinator not changed, BS1 coordinate transaction with help of BS2

MU inform the new BS of its coordinator


Dynamic methodThe role of coordinator moves with MU

The new BS informs all DBSs of the new coordinator

Mobile Transaction Model

Execution model based on ACID transaction framework:Execution Model with Reporting

TransactionTwo-Level Consistency ModelPro-Motion: Proactive management of

Mobile TransactionsPre-Write transaction Execution Model


MDS Transaction Management

Mobile Transaction execution.

DBS4

DBS1

DBS3

DBS2

T2(e4, e5)

MU2

MU1 T1(e1, e2, e3)MU3



Transaction fragments for distributed execution

Execution scenario: User issues transactions from

his/her MU and the final results comes back to the

same MU. The user transaction may not be

completely executed at the MU so it is fragmented

and distributed among database servers for

execution. This creates a Distributed mobile

execution.

Execution Model with Reporting Transaction

Based on Open-nested transaction A parent ACID transaction is set of

component transactions, where each component can be further fragmented in to a number of lower level components

Types of component transaction: Atomic transaction

It is an ACID transaction and can be compensated It can be commit before parent commit

Types of component transaction

Noncompensating component It is an atomic component but cannot be

compensated It delegate its updates to its parent transaction,

which handles the final commit Reporting transaction

Shares its update with the parent transaction at any time

Reports to other transaction by delegating some of its current result

Types of component transaction

Co-transactionLike reporting transaction but they can maintained

their execution state, suspend and resume This schema also identified component as

vital and none vital, a transaction commit if all vital component commit and all non vital component commit or successfully aborted

The reporting and Co-transaction components are responsible for all data modification

Execution Model with Reporting Transaction

Reporting component always execute on base station where co-transaction execute on MU

When a MU, running a co-transaction, moves and register with another BS, the corresponding reporting component also move to the BS and the communication between reporting and co-transaction remain active

Co-transaction get the most up-to-date values for data items

Reporting component receives update to install in the database

This model does not addressed location-dependent data processing issues.

Two-level consistency Model

A flexible execution model which considered data distribution as a part of the model

It created clusters of semantically related or closely located data which were dynamically configured

Data Cluster: A cluster is a set of data with a similar set of properties.

The model also allows mobile user to specify conditions for creating clusters

Useful for local computers who frequently access a well-defined set of data items

Mobile database MD={CL1,CL2, … , CLn}, CLi is a cluster.


Cluster consistency:There exist two kind of consistency:

Intra-cluster consistency: refers to consistency of data items of a cluster (local consistency)

Inter-cluster consistency: refers to data item consistency across the clusters similar to global consistency

The model defines weak transaction and strong transaction

Defines weak read and weak write and refers to the normal read as strict read and normal write strict write.


Strict write: it writes and leaves the database in a consistent state

Strict read: it read main database items which were written by the last strict writes

Weak write: Write local data (data in cluster). This data value becomes permanent only whit installed in the main database

Weak read: Reads local data (data in cluster) which is written by weak write


Weak transaction: a weak transaction is made up of only weak reads and writes

Strong transaction: a strict transaction is made up of only strict reads and writes

When a transaction is initiate with user, then it fragmented into a number of weak and strong transaction

Pro-Motion: Proactive management of Mobile Transactions

This model focuses on the issues introduced by disconnections and limited resources

The model manages the execution with the help of compacts and exploits object semantics for improving concurrency

Compacts: A compacts is an object which is composed of (a) cached data (b) cached data access method (c) information about the current state of the compact (d) rules of maintaining global consistency (e) obligations such as deadlines, and (f) an interface for the mobile unit to manage the compact

Compact object structure

Pro-motion Model

The server comes to an agreement with the mobile unit where a transaction is to be executed through a compact

MU provides its update through the compact The creation and removal of a compact is handled by

database server When the MU need data, then send the request to

server, after that, server creates a compact, initials with data and other necessary information, and send to MU

If there already exist compact then server update it.

Pro-motion Model

The transaction processing under Pro-Motion model has four basic step:Hoarding: the required compacts are prepared to

manage disconnected execution of transactionConnection execution: the MU connected to the

network and the compact manager processes transactions

Disconnected execution: the MU disconnected from the network, and compact manager processes transactions locally

Pro-motion Model

Resynchronization: the MU, after disconnected execution, is reconnected to the server and the compact agent reconciles local updates performed during disconnected execution with the server database

Mobile Transaction Model

HiCoMo: High commit Mobile transaction Model

Moflex Transaction ModelKangaroo Mobile Transaction modelMobilaction Model


The execution model is mainly for processing aggregate data stored in a data warehouse which resides in mobile units

Since the data warehouse resides in mobile units, HiCoMo transactions are always initiated on MU where they are processed in a disconnected mode

As a result transaction commitment are quite fast. The base database resides on the fixed network The structured of HiCoMo transaction is based on

nested transaction model.


It manipulated by transaction called base or source transactions. These transaction initiated at the fixed network.

This require that to install updates of HiCoMo transaction, they must be converted to source transactions.

This conversion is done by a Transaction Transformation Function

Moflex Transaction Model

Its based on Flexible transaction model, and a Moflex has 7 components and can be defined as Moflex transaction T = {M, S, F, D, H, J, G}

Moflex Transaction Model

If the parent subtransaction is compensable and processing location-dependent data, then the handoff rule forces the subtransaction to abort or restart in new cell. A restart can be split-restart where the value of the partial execution of the partial execution of the subtransaction in the last cell is preserved.

In the case of location-independent subtransaction, it further splits in finer subtransactions, the last subtransaction that occur in the last cell is free to commit.

Kangaroo Mobile Transaction Model

Presented which captured both data and the movement of mobile unit

The model based on a split transaction and enforce the ACID properties

A parent Kangaroo transaction, KT, is composed of a number of subtransaction, that each of them an ACID transaction, and called Joey Transaction (JT) and are local to a base station

Upon initiation of a Kangaroo transaction, a base station creates a JT for its execution

When the MU migrates to another cell, the BS of this cell takes control of the execution of this transaction

Kangaroo Mobile Transaction Model

A KT, when initiate by a MU, the initial BS immediately creates a JT with a unique identify and become responsible for its execution

When a MU encounters with handoff KT is split in two JTs, thus the mobility of a MU is captured by splitting a KT

This JTs executed sequentially, all subtransaction of JT1 are executed and committed before JT2’s

Mobileaction Model

A mobile transaction (MT) can be defined asTi is a triple <F, L, FLM>; whereF = {e1, e2, …, en} is a set of execution

fragments,L = {l1, l2, …, ln} is a set of locations, andFLM = {flm1, flm2, …, flmn} is a set of

fragment location mapping where j, flmi (ei) = li

Mobileaction Model

An execution fragment eij is a partial order eij = {j, j}

where

i = OSj {Ni} where OSj = kOjk, Ojk {read, write},

and Nj {AbortL, CommitL}.

For any Ojk and Ojl where Ojk = R(x) and Ojl = W(x) for

data object x, then either Ojk j Ojl or Ojl j Ojk.



In MDS a transaction may be fragmented and may

run at more than one nodes (MU and DBSs). An

efficient commit protocol is necessary. 2-phase

commit (2PC) or 3-phase commit (3PC) is no good

because of their generous messaging

requirement. A scheme which uses very few

messages, especially wireless, is desirable.

Transaction commit.



Transaction commit.

One possible scheme is “timeout” based

protocol.

Concept: MU and DBSs guarantee to complete

the execution of their fragments of a mobile

transaction within their predefined timeouts.

Thus, during processing no communication is

required. At the end of timeout, each node

commit their fragment independently.



Transaction commit.

Protocol: TCOT-Transaction Commit On Timeout

Requirements

Coordinator: Coordinates transaction commit

Home MU: Mobile Transaction (MT) originates here

Commit set: Nodes that process MT (MU + DBSs)

Timeout: Time period for executing a fragment



Activities of MUh:




Activities of CO:



Activities of DBs:



TCOT with Handoff:

Wireless information Broadcast


Data Dissemination ModeBroadcast Mode: in this mode the

broadcast server periodically broadcast most popular data on some wireless channel form which users can listen and, if necessary, download if required data. There is no uplink channel involved in this mode


Data Dissemination ModeOn-Demand Mode: this mode allows a

client to request specific data which is not available in the current broadcast or may never appear in the broadcast. The client send the query for require data through an uplink channel


Data Dissemination ModeHybrid Mode: in this mode, broadcast and

on-demand modes combined.


Pull processPull process is user (client) oriented, the user

assume that the desire in the wireless space, and he pulls it by tuning the channel.

Its user friendly and provides interactive capability to users for accessing the information through query

In wireless data dissemination platform, the pull approach is resource-intensive. Separate channel for sending the query and get the response from server are required.


Push processThe server broadcast data on one or multiple

channels. For example it can push weather info and traffic info on the separate channels.

Client depending upon their data requirements, tune the appropriate channel

Client can not send the specific query to the serverMessage indexing can be used for speed up

broadcast search


Broadcast Disk


Broadcast Schedule

Any Questions?

mds_khazaei

Documents

communication device

estabiishing communication

multipoint communication

setup communication

communication equipment

wireless system

mobiie database system

mobile device