Transaction processing, techniques

International Journal on Computational Sciences & Applications (IJCSA) Vol.5, No.1, February 2015
1
TRANSACTION PROCESSING, TECHNIQUES
IN MOBILE DATABASE:
An OVERVIEW
Adil Mobarek1
,Siddig Abdelrhman2
,Areege Abdel-Mutal3
,Sara Adam4
,Nawal
Elbadri5
,Tarig Mohammed Ahmed6
Computer Science Department, Khartoum University, Khartoum,Sudan
ABSTRACT
The advancement in mobile technology and wireless network increase the using of mobile device in database
driven application, these application require high reliability and availability due to nature inheritance of
mobile environment, transaction is the center component in database systems, In this paper we present
useful work done in mobile transaction, we show the mobile database environment and overview a lot of
proposed model of mobile transaction and show many techniques used to enhance transaction execution.
KEYWORD
Mobile Transaction, Reference Architecture, Mobile Database, Mobile Host, Concurrency Control
INTRODUCTION
The transaction is program in execution in which each write-set satisfy the ACID properties [11], It
can be described as a sequence of operation that form a single of logical unit of work, if a
transaction takes place in mobile environment is called mobile transaction [12]
There are a lot of limitation in mobile environment introduce a lot of challenges to manage the
mobile transaction, the main challenges of mobile transaction management come from the mobility
of mobile host and the limitation in wireless bandwidth, the bandwidth available is very low
typically in the range of 10 Kb/s in case of cellular links and 2 Mb/s in the case of an infra red link.
[16]
These limitations make traditional transaction management concepts inadequate, a lot of transaction
models and concurrency control mechanisms are proposed to full fit with the mobile environment,
in this paper we overview of most importance works in this area.
This paper organized in ten sections, in Section 2,3 we discuss the environment of mobile database
and overview transaction models in mobile database, Section 4,5 show the concurrency control
mechanism used in mobile environment and mobile transaction commitment, Section 6,7 show
replication and synchronization in mobile database and caching in mobile database ,section 8 show
the recover in mobile database , lastly we show the conclusion and discussion in section 9,10.

2
2.OVERVIEW OF MOBILE DATABASE ENVIROMENT
Traditional database is processed by immobile processing unit both the processing unit and their
users are immobile at the time of data processing, In mobile database the user communicate
without wire by using the wireless network communication in which the communication links is
established through radio frequency (RF), the data transmitted using the range of radio frequency
called band, to describe the size of the band the term bandwidth is used. [17]
There is no radio frequency which can carry the data to long distance [17], to overcome this
limitation the cellular system use the handoff operation, normally occurs when the mobile is
moving, in which the communication cannel between the mobile host and old base station is
braked and the mobile take new channel to communicate with the new base station. Bellow is
the reference architecture of mobile database:
Here are the roots of limitations in mobile database system:
- The strength of radio channel is became week when the mobile user move in cell boundary
- When the mobile user move from on cell to neighbor cell, the handoff operation take
place, the hand off cannot be predictive so it introduce new challenge in mobile database
- The bandwidth of the cell is limited, so limited number of mobile user can connect with
mobile database, so the problem will occurs when so many user enter the cell shadow.
- The mobile device is also source of challenges; the battery life time is small, and the
limitation of processing power and storage capacity.
These limitations in mobile database environment make the mobile transaction have special
characteristics: [9]
- Transaction in mobile takes long-lived due to the frequent disconnection also due to data
and users also move from place to others
- Transaction computations can be split in sets some executed in mobile host and other in
mobile service station.
- The roles of Mobile Service Stations (MSS) are computation and communications.
- Accuracy mobile transaction should support and deal with concurrency, recovery,
disconnections and mutual consistency of the replicated data objects, below we show the
architecture of mobile database.
Vijay Kumar presents Reference architecture (Figure.1) for mobile database system. The
architecture consist of two types of computer, fixed host and mobile host, the fixed host connect
throw the wired network and mobile unit connect with wireless network using the base station
(BS) as access point, the BS act as interface between the mobile host and the fixed host, the base
station and fixed host are connected thought a high speed wired network, the component in this
model which control the base station is called the base station controller
BSC (base station controller), the HLR (home location register) is small database contains
information about the network subscribers, the MSC (Mobile Switching Center) and PSTN (public
switching telephony network) are used as gateway to international or others network, the fixed and
mobile host has full access to DB (database) using the DBS (database system) [17], When an MU
leaves a cell boundary, a hand-off protocol is used to transfer the responsibility for mobile
transaction and data support to the BS of the new cell. This hand-off involves break the connection
with old BS and establish new connection to connection with neighbored BS. It may also involve
migration of in progress transactions and database states from one BS to another. [4]

3
Figure1. Reference Architecture [17]
Apart from the Reference Architecture for Mobile database, there are three types of architectures
discussed in the literature namely Client/Server architecture, Server/Agent/Client architecture and
Peer-Peer architecture.
3. MOBILE TRANSACTION MODELS
Specific characteristics of mobile environments (e.g. variable bandwidth, disconnections, and
limited resources on mobile hosts) make traditional transaction management techniques no longer
appropriate [5]. Several models for mobile transactions have been proposed, in this section we
overview several proposed models of mobile transactions
3.1. Kangaroo Transaction Model
The nature of mobile environment allow mobile host to move from base station to another, This
model solve the problem of mobile host movement during the execution of the transaction, the
mobile transaction start from the mobile host and finally executed at the MDBS on the wired
network, the Kangaroo transaction propose to implements data access agent on the top of existing
global transaction manager, the agent is placed in all base stations, the role of this agent to
manage mobile transaction and mobile host movement, when the mobile move from base
station to another the coordination of mobile transaction also moves. [5]
A Kangaroo transaction has a unique identification number assembled from the base station
number and unique sequence number within that base station, when handoff operation is
executed due to mobile host movement, the control of the Kangaroo transaction move to a new
DAA (Data Access Agent) at another base station. [10]

4
3.2. Clustering transaction model
This model add big enhancement in mobile environment that suffer from disconnection variation by
adding replication scheme, the database dynamically divided into clusters, each one group together
related data, maintains two copy to all object in database, on copy called strict version must be
globally consistence, the other version called weak version can tolerate some of globally
inconsistence but must be locally consistence, when the mobile host is weakly connected or
disconnect, the update will applied on weak version on database object, so the degree of
inconsistency between week and strict version very depend on the availability of network
bandwidth among clusters. [5]
The mobile transaction start on mobile host, this model divide the transaction into two types, strict
transaction, which access strict version of data object, and executed when the hosts are strong
connected, the other type of transaction is week transaction, it access only the weak version, and
executed only when the hosts are weakly connected or disconnected, the strict transaction execute
the standard read and write( strict operations) whereas the weak transaction execute the weak
operation, and at the reconnect a synchronization process executed on the database server lead the
database to global consistency.[ 5 ]
3.3.Two-Tier transaction model is
The basic idea of two-tier transaction model is first to allow s the transactions to run locally on MU
which update local data, and at the time of reconnection all transactions are re-executed at servers
as base transactions (BT) [8].
This transaction model use the replication mechanism, the ager replication mechanism is not
effective in mobile environment, so this model use the lazy replication mechanism; there is master
and replicated version for each data [5], two types of transactions are used, Base transactions which
access master versions of data and tentative transaction which executed when the mobile transaction
in disconnection mode, the transaction execute the update on mobile host, when the mobile host
reconnect, tentative transactions are converted to base transaction that are executed on master copy.
[9]
3.4. Pro-motion transaction model
Pro-motion is mobile transaction processing system that support the disconnection mode, it enhance
the cashing mechanism used on mobile host to make possible local transaction processing in
consistence mode. [5], Pro-motion transaction model use the concepts of nested spilt transaction, it
is one example of open nesting, which relaxes the top-level atomicity restriction of closed nested
transactions where an open nested transaction allows its partial results to be observed outside the
transaction. [15]
3.5. Twin Transaction Model
Defines a transaction execution mechanism which achieve the needing of connecting and
disconnecting modes of operation, this model uses resynchronization mechanism to achieve the
consistence state after the mobile host is connected to the network,
3.6. PMTM
New theoretical mobile web transaction model is proposed by Zhengwei et al. called PMTM (P
system-based Mobile Transaction Model) to be suited the behavior of mobile transactions,This

5
model use two transition rules, Membrane rules and Object rules, The Object rule describes the
transitions in membranes whereas the Membrane rule defines the structural modification of
membranes. [17]
3.7.Adaptable Mobile Transaction Model
Allows defending transactions to have many execution alternatives related to particular context, this
model is full fit transaction execution to context diffraction, in this model preserve conflict
serializability but relax the Atomicity and Isolation, this model improves the commit possibilities
and allows selecting the way transactions will be executed according to their costs.[17]
3.8. Shadow Paging Technique
Mobile-Shadow technique for handling mobile transaction processing and disconnection, is
proposed by Osman Mohammed Hegazy et al. in [11], it’s new enhancement on shadow
paging technique, M-Shadow uses the idea of actionability, which differentiates the
actions to be taken during the transaction's validation phase according to the types of
affected attributes.
3.9. Surrogate Object Based Mobile Transaction Model
New transaction scheme (an enhancement of kangaroo model) represented by Ravimaran
S. et al. in [14],The main issue is supporting data cashing at surrogate object for reducing
data seeking and enhance the database operations among mobile transactions at vary
mobile hosts in mobile environment. The tested results demonstrate that there is a
significant reduction in wireless access and abort can be handled with the proposed model.
3.10. Connection Fault-Tolerant Model
For mobile environment which minimize the blocking time of resources at the fixed devices
supports fast recovery from connection failures due to mobility of mobile devices and increases the
number of committed mobile transactions.
3.11. MDSTPM(Multidatabase Transaction Processing Manager architecture)
It is framework to improve the transaction submissions from MHs in an heterogeneous
multidatabase environment, this model concerns on implementation of the Message and Queuing
facility which manages the exchanges between MHs and the wired multidatabase when mobile
disconnected.
MDSTPM must be at each host (FH) on top of existing local DBMS, Local DBMSs responsible for
local processing, whereas MDSTPM coordinates the execution of global transactions, it generates
scheduling and coordinates commits.
FH coordinator is designed in advance, to manage transactions on behalf of the MH when it
disconnected. In this model all ACID properties are enforced on all sites of wired network. [17]

6
4.CONCURRENCY CONTROL MECHANISM USED IN MOBILE
DATABASE
The poor quality of services provided by a mobile network can seriously increase the overheads and
affect the effectiveness of a concurrency control protocol in resolving data conflicts as the
transactions now require a longer time for completion [2]. Current multidatabase concurrency
control schemes do not address the limited bandwidth and frequent disconnection associated with
wireless networks. [6] The concurrent access anomalies occur when many mobile hosts need to
access the same data items. To preserve Isolation property some of Concurrency control
mechanisms are proposed in literature, in this section we will show brief overview of the most
important methods for increasing concurrency between transactions and make some compactions of
these methods.
The first method to increase concurrency control is called Pessimistic concurrency control; in this
method the locking protocol assume that the data used by one transaction may be needed by others
transaction therefore it is better to lock the data. In this method, locking protocol implements the
serializability by using lock, transaction lock the data it uses to prevent other transaction from
access it is data. To optimize the pessimistic concurrency control it is better to use shared lock if the
transaction jus need to read the data item, and use excusive lock if the transaction need to
read/update the data item, all the lock used by the transaction should not released until the
transaction is commit. To enable the program use short duration of the lock the transaction designer
may divide the transaction into sub transactions that manage the locks independency of parent
transaction, The disadvantage of this method are the relax ASID is difficult to use, and the long
duration of the lock may case deadlock [7].
The second method for increasing concurrency control is Optimistic concurrency control, in this
mechanism the data items are not locked and can be updated by more than one mobile host at the
same time. [13], the transaction assumes that data will be accessed by the transaction is not needed
by other concurrent transaction, so the transaction doesn’t need to use the lock protocol, at the
commit time the transaction chick if a conflict has occurred, if there are conflict at commit time
offending transaction is restarted, if no conflict all updates of transactions are written to database
and the transaction is committed. The Optimistic concurrency control doesn’t use the lock as all
transaction has access to all data until commit time, if the transaction is divided into subtransaction
it is possible to increase concurrency as the space of the tie in which conflict may occurs reduced,
also if we use subtransaction the relax ACID properties should be used [7].
In an optimistic strategy, whenever multiple sites request for same data items, they are allowed to
read the data item thus tolerating conflicts to occur. However at the time of global commit, conflict
resolution strategy is applied to get back the system into consistent state. In mobile database
environment, the global commit operation may increase the uplink bandwidth as the mobile host
has to initiate the transactions again and again. [3]
The third method for increasing concurrency between transactions is called Low Isolation Level;
the author defined the isolation level of a transaction as the degree of interference tolerated on
the part of concurrent transactions [7].
When we use low isolations level more concurrency is allowed and more different isolation
anomalous may occurs in the transactions, it is useful to choose low isolation level as
possible in the application in order to archive high performance and availability.

7
The table below show the anomalies allowed by the different isolation levels of SQL. [7]
Table 1: anomalies allowed by the different isolation levels
The fourth method for increasing concurrency between transactions is Short Duration
Locks, which are the locks that are released as soon as possible, example data will not be locked
across locations, this method is very useful in distributed systems, in which the transactions often
consist of a root transaction in a client location that manage serial sup transaction on one or more
servers, each time the root transaction receive new info from the user, it start new subtransaction in
the server each time until the transaction is finished. [7]
mobile transaction is long-lived transaction as the user may b relatively slow in making his/her
input, and others concurrent users don’t want to wait for the first user to finish his/her job before
they can access the same data, this is special case if the first user locks info to finish his/her job,
therefore short duration lock are often used to increase concurrency, in the system with relatively
many updates like ERP-Systems and E-commerce Systems low isolation level cannot solve the
availability problem as all update lock must be exclusive, in such situation is recommended to use
short duration locks. [7]
5.MOBILE TRANSACTION COMMENTMINT
The commitment of mobile transaction can be affected by Mobile mobility and other limitation of
mobile environment,the common limitation are mobile battery power, disconnection from its base
station, disk space limitation, physical abuse, accident, and unpredictable handoff. There are special
protocols used for commit mobile Transaction to full fit the inheritance limitation of mobile
environment therefore, some commit protocol is necessary for their termination. Conventional
commit protocols such as 2PC, 3PC, etc., could be molded to work with MDS. [12]
6. REPLICATION AND SYNCHRONIZATION IN MOBILE
DATABASE
Data replication is used in mobile database to enhance response time of access the data, to achieve
data availability and to reach for full offline functionality of data. There are two types of
replications Eager and Lazy Replications; In Eager replication, grounded in atomic of transaction
i.e. the transactions must applied to all nodes.
In Lazy replication the update applied in master node and the others replicas need different
transaction for each node.
SQL isolation
level
Isolation anomalies
Dirty read Non-
repeatable
read
phantoms
SERIALIZABLE No No No
REPETABLE
READ
No No Yes
RED
COMMITED
No Yes Yes
READ
UNCOMMITED
Yes Yes Yes

8
An Optimistic Client Centric Replication for relational database is proposed to full fit the
characterize of mobile environment for example disconnection, weak connectivity, limited
bandwidth and poor resources, using this protocol led to reduce the connection between the server
and client by transmitting only the modified data.
There are two type of policies categorized into Pessimistic and Optimistic, Pessimistic way is made
the update in single or many replicas by put restriction or lock on the replica to prevent the access in
updating time, this way led to guarantee consistency and is called one-copy equivalence. The
Pessimistic is not possible now in mobile due to require continuous communication and less
resource availability of the mobile devices.
Optimistic way in this way the operations can be executed in any node but at time of reconnection
the propagate will take place and therefore the data is inconsistency not like pessimistic way. There
are two levels of synchronization techniques are proposed in literature [1], Data Level
Synchronization (DLS) and Transaction Level Synchronization (TLS), In DLS technique the basic
unit of synchronization is data whereas the TLS it depend on transaction as basic unit, Rely on the
simply and efficiency the DLS in most suitable option but it cannot guarantee for atomicity of
transaction and consistency of the database, on the other hand TLS guarantees both atomicity and
consistency ,thus the TLS is better appropriate technique for synchronization of mobile database.
Exo-leasing is presented in this grounded on grantee synchronization without need to run any code
in the servers. In this technique grantee synchronization code executed in the client and allows a
disconnected client to obtain grantee reservation from another disconnected client. [3]
7.CACHING IN MOBILE DATABASE
The major purpose of cashing is enhancing data availability and performance of accessing to data
from central database stored in severs. Performing local queries in cache of mobile device and latter
update data in central data base. Cashing technique increase queries response time and it is very
efficient when temporal locality exists in the access patterns between a query and its proceeding
queries. Whenever a query is issued, the client cache manager checks its own cache, this technique
is the key to handle problem of disconnection in mobile computing.
8. RECOVER IN MOBILE DATABASE
Recovery is difficult task in mobile database due to the mobility of processing node, fixed network
failure, bandwidth limitation, and site failure and handoff problems. As a solution for these
problems two types of protocol are used for saving the execution state of a mobile application,
namely Checkpoint and Message Logging, the mobile application uses these protocols when failure
occurs for rolling back to the last saved consistence state, and restart execution with recovery
guarantees. Checkpoint and log information are stored at the base stations since Mobile Host disk
storage is not stable.[12] In the literature three approach of Checkpoint are found; Uncoordinated
Checkpoint, Coordinated Checkpoint and Communication-Induced Checkpoint. In Uncoordinated
Checkpointing each participating process takes its checkpoints independently, in this approach the
system is prone to cascaded rollback, in Coordinated Checkpointing protocol each processes
coordinate their checkpoints to form a System-wide consistent state. In this approach, if any failure
occurs, the system state can be restored to a consistent set of checkpoints and avoiding the rollback
propagation. The Communication-Induced Checkpointing approach is force every process to take
checkpoints based on information piggybacked on the application messages it receives from other
processes.

9
9.CONCLUSION
The mobility is very attractive in wide range of application, especially in database application, the
inheritance limitation of wireless add difficulty to manage the transactions in mobile Database,
Hence this field motivates a lot of researcher to design techniques and algorithms to suit the mobile
transaction behavior. This paper has provided an overview of the state-of-the-art in mobile
transaction, we overview the mobile environment and show its main limitations, a lot of mobile
transaction models are proposed in literature to cope the MH movement and others wireless
limitations, every model uses special techniques to overcome these limitation, for example two-tire
transaction model uses replication mechanism, pro-motion transaction model support disconnection
by enhancing the cashing mechanism, some of models support recovery from failure like
connection fault tolerant model, the MDSTM model supports transaction submission for MH in
heterogonous multi database environment.
To preserve Isolation properties of mobile transaction some of concurrency control mechanism are
proposed in literature, we show four methods to increase concurrency, which are pessimistic
concurrency control, optimistic concurrency control, low isolation level and short duration lock,
every mechanism has strength and weakness when used in special mobile application. The
mobility of MH and other wireless limitation affects the commitment of mobile transactions, there
are special protocols used in mobile environment for mobile transaction commitment. Data
replication is used in mobile database to enhance response time of access the data; we show two
types of replication, ager replication and lazy replication, also we show two levels of
synchronization techniques which are proposed in literature, Data Level Synchronization (DLS)
and Transaction Level Synchronization (TLS).
In order to enhance data availability and performance of accessing data from central database
cashing technique is used. The recovery is difficult task in mobile database due to the mobility of
processing node, fixed network failure, bandwidth limitation, and site failure and handoff
problems, we show to protocols to save the state of execution of a mobile application, namely
Checkpoint and Message Logging.
The advancement of mobile device will led to using mobile in sensitive application like electronic
payment and mobile banking, which require high dependability and security, will need more
enhancements in mobile transaction management to suite the requirement of these applications, this
survey show a lot of proposed models for mobile transaction, but still there is no mobile transaction
model solve all wireless limitation, some mobile transition model solve the movement behavior of
MH like kangaroo transaction model, others models solve the disconnection problem, this means
that some model has good performance in some application, some mobile application require high
database consistency, and some are not, but need high availability of data, the advancement of
mobile hardware may give the ability to propose more enhanced mechanism for transaction
processing, for example the replication mechanism need big storage capacity of MH.

10
REFERENCES
[1] Ali Karami1 , Ahmad Baraani-Dastjerdi, “A CONCURRENCY CONTROL METHOD BASED ON
COMMITMENT ORDERING IN MOBILE DATABASES”, International Journal of Database
Management Systems ( IJDMS ) Vol.3, No.4, November 2011
[2] Karami, Ali, and Ahmad Baraani-Dastjerdi. "A CONCURRENCY CONTROL METHOD BASED ON
COMMITMENT ORDERING IN MOBILE DATABASES." International Journal of Database
Management Systems 3.4 (2011).
[3] Salman Abdul et al. “CONCURRENCY CONTROL IN MOBILE ENVIRONMENTS: ISSUES &
CHALLENGES”, International Journal of Database Management Systems ( IJDMS ) Vol.3, No.4,
November 2011
[4] S.K. Madria et al. “Mobile data and transaction management” Information Sciences 141, April 2001
[5] P.S Alvarado “Survey of Mobile Transactions “, Distributed and Parallel Databases, July, 2003
[6] James B. Lim and A.R. Hurson, “Transaction Processing in Mobile, Heterogeneous Database Systems”,
IEEE
TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, VOL. 14, NO. 6, 2002
[7] Lars Frank, “Databases and Applications with Relaxed ACID Properties”, Doctoral Dissertation,
Copenhagen business School, Samfundslitteratur 2008.
[8] N. Prabhu, “Concurrency Control in Mobile Database Systems”, IEEE 2004
[9] Malasowe Bridget and S. A Idowu , “Current State of the Art in Mobile Transaction Models: A survey
“, Nigeria IJASCSE Volume 3, Issue 1, 2014
[10] P.K Panda et al. “Review of Some Transaction Models used in Mobile Databases”, International
Journal of Instrumentation, Control & Automation (IJICA), Volume 1, Issue 1, 2011
[11] Hegazy, Osman Mohammed, Ali Hamed El Bastawissy, and Romani Farid Ibrahim. "Handling Mobile
Transactions with Disconnections Using a Mobile-Shadow." Cairo University, Egypt. 2008.
[12] D. Roselin Selvarani and Dr. T. N. Ravi, “A SURVEY ON DATA AND TRANSACTION
MANAGEMENT IN MOBILE DATABASES” IJDMS Vol.4, No.5, October 2012
[13] S. Abdul MoizI et al. “Commit Protocols in Mobile Environments: Design & Implementation” , JDMS
Vol.2, No.3, August 2010
[14] Ravimaran S. et al. “Reliable and Fault Tolerant Mobile Transaction Paradigm using Surrogate Object”,
International Journal of Future Computer and Communication, Vol. 1, No. 1, June 2012
[15] Seydim, Ayse Yasemin. "An Overview of Transaction models in mobile Environments." Dallas:
Departament od Computer Science and Engeneering Southen Methodist University (1999).
[16] Helal, Abdelsalam, et al. “A Survey of Mobile Transaction Models”, Purdue University- Department of
Computer Science 1996.
[17] Kumar, Vijay. “Mobile database systems”. John Wiley & Sons, 2006.

Transaction processing, techniques

More Related Content

What's hot (18)

Viewers also liked (20)

Similar to Transaction processing, techniques (20)

Recently uploaded (20)

Transaction processing, techniques