nam report

7/18/2019 Nam Report

http://slidepdf.com/reader/full/nam-report 1/10

NAM

Mobile Computing & Networking

Submitted to: Submitted by:

Ms. Dimple Chawla Mohit Vats (11!

("a#ulty $ D%S!



Abstract

Advances in wireless networking have prompted a new concept of computing, called mobile computing in which users carrying portable deviceshave access to a shared infrastructure, independent of their physical

location. This provides exible communication between people and ideally!continuous access to networked services. Mobile computing is revolutioni"ingthe way computers are used and in the coming years this will become evenmore perceptible although many of the devices themselves will becomesmaller or even invisible such as sensors! to users.

This essay attempts to give an insight into mobile computing, in particular,software design issues models, algorithms and in particular middleware! areconsidered. #uch issues arise from the need for wireless networking, theability to change location and the need for unencumbered portability as wellinsuring security standards comparable to that found is distributed systems

or central systems.

1. Introduction

$n the last %& years, the advent of mobile phones as well as laptops hasdramatically increased the availability of mobile devices to businesses andhome users. More recently, smaller portable devices such as '(As andespecially embedded devices e.g. washing machines, sensors! have slowly

changed the way humans live and think of computers.

)omputing is drifting away from *ust being concentrated on computers andrelates moreand more towards society, its people and its infrastructures. This is particulartrue wheresensors are being developed to be so minute that they are literallyembedded in clothing and even humans.

Mobile computing is associated with the mobility of hardware, data andsoftware in computer applications. The study of this new area of computing

has prompted the need to rethink carefully about the way in which mobilenetwork and systems are conceived. +ven though mobile and traditionaldistributed systems may appear to be closely related, there are a number of factors that dierentiate the two, especially in terms of type of device-xedmobile!, network connection permanentintermittent! and executioncontext staticdynamic!



2. Diferent types o Mobile Systems

$n many ways, mobile computing has several characteristics reminiscent of distributedsystems. $n order to understand mobile systems, one must -rst understand

where thesimilarities and the dierences of distributed and mobile systems lie. Thefollowing section is an explanation of the dierent types of distributedsystems ranging from the traditional type to nomadic, ad/hoc and -nallyubi0uitous ones.

Traditional Distributed System

Traditional distributed systems consist of a collection of -xed hosts that arethemselves

attached to a network1 if hosts are disconnected from the network this isconsidered to be abnormal whereas in a mobile system this is 0uite thenorm. These hosts are fxed and are usually very powerful machines with fastprocessors and large amount of memory. The bandwidth in traditionalsystems is very high too.

2urthermore, the execution context is said to be static as opposed to adynamic contextwhereby host *oin and leave the network fre0uently. $n a traditional system,location rarely changes as well and hosts are much less likely to be added ordeleted from the network.

Traditional distributed systems also need to guarantee non/functionalre0uirements such as scalability accommodate a higher load at some timein the future!, openness possibility to extend and modify the system easily!,heterogeneity integration of components written using dierentprogramming languages, running on dierent operating systems, executingon dierent hardware platforms!, fault/tolerance recover from faults withouthalting the whole system! and -nally resource/sharing some form of accesscontrol!.

Nomadic Distributed System

This kind of system is composed of a set of mobile devices and a coreinfrastructure with -xed and wired nodes. Mobile devices move from locationto location, while maintaining a connection to the -xed network. There areproblems that arise from such shifts in location. The mobile host has a home$' address and thus any packets sent to the mobile host will be delivered tothe home network and not the foreign network where the mobile host is



currently located. #uch problem can be solved by forwarding packets to theforeign network with the help of Mobile $'. Nevertheless, Mobile $' alsosuers from e3ciency routing issues!, 4o#, security authentication of mobile host at foreign network and end/to/end security re0uired! andwireless access reduced capacity! problems.

These systems are susceptible to the uncertainty of location, a repeated lackof connections and the migration into dierent physical and logicalenvironments while operating. 5owever, compared to ad/hoc networks,nomadic systems still have comparatively reliable connections and servicessince most of these are actually supported by the -xed infrastructure6backbone7! of the network.

The non/functional re0uirements mainly dier, compared to the traditionaldistributed systems, in the heterogeneity aected by the presence of both-xed and mobile devices across the network as well as the variations in

technologies e.g.8 wireless!!, resource sharing must take into accountdierent issues when the resources need to be discovered! and faulttolerance of the system considered to be 0uite the norm!. 4uality andprovision of these resources must be carefully considered too.

Ad-Hoc Mobile Distributed System

Ad/hoc distributed systems are possibly the only type of network that comesclose to mobile networks in the sense that every node is literally mobile. $t isthese networks that are very much seen as the systems of the future,

whereby hosts are connected to the network through high/variable 0ualitylinks e.g.8 from 9'# to broadband connection! and executed in an extremelydynamic environment.

A/hoc systems do not have any -xed infrastructure which diers them bothfrom traditional and nomadic distributed systems. $n fact, ad/hoc networksmay come together as needed, not necessarily with any assistance from theexisting e.g.8 $nternet! infrastructure. :hen nodes are detached from the-xedmobile network they may evolve independently and groups of hostsopportunistically form 6clusters7 of mini/networks. The speed and ease of deployment make ad/hoc networks highly desirable.

These kinds of systems are extremely useful in conditions where theinfrastructure is absent, impractical to establish or even expensive to builde.g.8 military applications, high terrain uses, and emergency relief operations!. 5owever, being a relatively new technology, this -eld of networkingdemands a lot of research to be done to improve it, especially its non/



functional re0uirements as well as algorithms for routing protocols e.g.8distance vector and dynamic source routing algorithms!.

#ecurity threats have to be dealt even more cautiously in ad/hoc networks.(esigning secure key distribution in an ad/hoc network might be an

extremely hard task. Any reliance on a certi-cate of authority is not trivial atall, for the same reasons that reliance in any centrali"ed authority isproblematic. Additional problems include the increased packet si"es re0uiredby authentication extensions, unicastmulticast routing, 4uality of #ervicesupport and power aware routing. 2urthermore, due to the limitedtransmission range of wireless network interfaces, multiple hops may beneeded to exchange data between nodes in the network c.f. MAN+T!.

. T!eory in Mobile "omputin#

This section is only an introductory section so it will not go into any detailother than state what the current trend of search in these two -elds relatedto mobile computing are8Models

Models permit the precise description of existing languages and systemsemantics. $n fact, they enable the formal reasoning about the correctness of

such semantics. Models are very much used to emphasi"e parallels anddistinctions among various forms of mobility logical and physical! and areconcerned with the formulation of appropriate abstractions useful inspeci-cation and evaluation of such mobile systems.

Models are mainly concerned with the characteristics of mobile units such asthe unit of mobility who is allowed to move!, its location where a mobile unit ispositioned in pace! and its context determined by the current location of mobile units!. There are many existing models and many more are still inresearch8

• ;andom mobility models!• Markovian model

• +xponential )orrelated ;andom Model• Nomadic )ommunity Model



Al#orit!ms

The current algorithms applied reect the assumptions that are made aboutthe underlying system. <nfortunately, many of these assumptions are notsuited for current

algorithms for mobile systems. Mobile algorithms are obliged to treat inmuch detail space and coordination of mobile systems. $n particular,algorithms have to carefully take into consideration location changes, thefre0uency of disconnection, power limitations and the dynamic changes inthe connectivity pattern of mobile systems. This -eld of theory is in factspread among a vast spectrum of research due to the large diversity of mobile systems.

$. Middle%are

A lot of research has been made in recent years into the translation of traditional middleware into that of mobile distributed systems. 5owever, thisis not as easy as researchers -rst thought due to the dierences betweentraditional and mobile systems stated previously. Traditionally, middlewarefor physical mobility= has been application centred e.g.8 >ayou system!.Nevertheless, this approach is not suitable for a generic form of mobilecomputing and therefore a general purpose middleware becomes anecessity. 5iding mobility is increasingly more di3cult or even meaningful,thus a new core of abstractions that extend distributed middleware withsupport to mobility must be fully researched.

Another reason for the so/called 6lack7 of a suitable mobile systemmiddleware is that traditional distributed systems have been around for over?& years whereas mobile systems are a new technology which has beenaround at most %& years. This means that this new -eld is still very muchinto research.

Middleware for nomadic and ad/hoc mobile distributed systems has a set of comparablecharacteristics that inuence how the middleware should in fact behave.Mobile devicesre0uire light computational load 1 existing middleware for heavycomputational load such as that found in traditional distributed systemscannot be applied. The intermittent connection nature of mobile systemsalso re0uires an asynchronous form of communication. Additionally, unlike-xed distributed systems, mobile systems execute in an extremely dynamiccontext which in turn necessitates that devices be aware at all time of theirenvironment e.g.8 type of connection they are switching to!.



Middleware for mobile distributed systems is split into dierent researchareas such ascontext/aware middleware principle o Reection!@ location/awaremiddleware e.g.8 Nexus!@ data sharing/oriented middleware e.g.8 Bayou,Coda, Odyssey !@ tuple space/based middleware e.g.8 Lime, TSpaces!. This

paper will only have a brief look at the tuple spacebased approach and, inparticular, ime will be used as an example.

Tuple Space Middle%are

The classical de-nition of a tuple space is that of a shared associativememory consisting of a collection of tagged data records tuples!B. Tuplesmay be created and placed in the tuple space and they can be accessconcurrently by several processes with blocking primitives. $n a mobilecontext a completely asynchronous de/couple communication! and

decoupled model is suitable and thus tuple space middleware is appropriate. Tuple spaces neednCt depend either on the machine or platform in which theyare running.

Tuple spaces have proven to be suitable for mobile computing because of thedynamiccontext nature migration and connectivity patterns! of mobile systems. Theyare an attractive approach for coordinating mobile units across a mobilecomputing environment.

&IM'

ime &inda In a Mobile 'nvironment! is a *ava/based middleware thatprovides a coordination layer that can be exploited successfully for designingapplications that exhibit either logical or physical mobility / or both. Thefundamental design criteria underlying ime comes from the reali"ation thatthe de-ning problem of mobile computing is dealing with, and exploiting, adynamically changing context. To achieve its goal, ime borrows and adaptsthe communication model made popular by indaD.

ime provides coordination among processes Lime aents! via a shared

memory mechanism. :hen processes use ime to coordinate, no messagesare explicitly sent to other processes. All communication occurs throughaccess to a shared medium, namely the tuple space.

The dierence between inda and ime is that the former breaks up thetuple space of the latter into several tuples spaces. +ach one of these tuplespaces is associated to a mobile unitE and every mobile unit has access to aninterace tuple space that is constantly and exclusively attached to that unit



and transferred along with it when movement occurs. The tuple space thatcan be accessed through this interface is thus shared by construction and istemporary because its content changes according to the movement of mobile units.

ime is a very powerful tuple space/based middleware since it promotes thereduction of details of mobility transparency ! and the distribution tochanges to what is perceived as the local tuple space. This allows designersof mobile systems to be alleviated from the burden of constantly maintaininga view of the context consistent with changes in the con-guration of thesystem. 5owever, transparency could lead to an oversimpli-cation of asystem. $n some cases, where there is a need to achieve a high amount of context awareness, this may not be favorable.

Nonetheless, ime oers several 6solutions7 such as making informationabout the system con-guration available via ime#ystem%&@ allow actions to

be taken in response to a change in the con-guration of the system bysetting reactions on the tuple space@ and broaden inda operations withtuple location parameters that permit to function on dierent pro*ections of the transiently collective tuple space.

(. )SM Hando*er Type

There are dierent categories of 9#M handover which involves dierent partsof the 9#M network. )hanging cells within the same >T# is not complicatedas the changing of the cell belonging to dierent M#). There are mainly two

reasons for this kind of handover. The mobile station moves out of the rangestation or the antenna of >T# respectively. #econdly the wire infrastructurethe M#) or the >#) may decide that the tra3c in one cell is too high andmove some to other cells with lower load. These are the main reasons thatinitiate dierent kinds of handover. 2ollowing are the dierent kinds of handover and their details 8

Intra-cell +TS Hando*er

The terms intra/cell and intra >T# handover are used both for fre0uencychange. There is a slight between them but usually they are considered the

same. The term intra/cell handover in not real as it deals with the fre0uencychange of a going call. The fre0uency change occur when the 0uality of thecommunication link degrading and the measurements of the neighboringcells better than the current cell. $n this situation the >#) which controls the>T# serving the M#) order the M#) and >T# to switch to another fre0uencywhich oers better communication link for the call. The communication linkdegradation is caused by the interference as the neighboring cell using thesame fre0uencies and its better to try another channel. $n the intra >T#



handover cell involved are synchroni"ed. $n the synchroni"ation the M#)sends four 5N(FA)) messages to the serving >T#. The 5N(FA)) message isonly one byte long and contains the handover reference. The >#) send the5N(F)M( to initiate the intra >T# handover. The >#) 5N(F)M( the time slotand the new channel. The connection is established between the >#) and

M#) by exchanging these messages. After these messages exchanged M#)receives 5N(F';+2 message that the handover is performed by the >#) andthen >#) re0uests >T# to release the resources that is no longer used. Thissynchroni"ed handover saves resources and faster than non synchroni"edhandovers.

>#) handover is entirely carried out by the >#), but the M#) is noti-ed whenthe handover has taken place. $f the targeted cell is in dierent location areathen the M#) needs to perform the location updates procedure after the call.$n the intra/>#) handover both synchroni"ed and non synchroni"ed handoverare possible. The -gure G shows the intra/>#) situation.

$n the intra/>#) handover the >#) send 5N(F)M( message which containthe time slot, the fre0uency of the new channel and how M#) shall identifyitself on the new channel. And at the same time >T# send '5H#F$N2Imessage and wait for the #A>M #et Asynchronous >alance Mode! from theM#). :hen the M#) receives '5H#F$N2I message then send #A>M in orderto establish A'( for mobile. :hen the >T# receives the #A>M it sends anempty as an acknowledgement to the >#). Now the M#) only receive theinformation that the handover was performed by the >#) and >T# releasesthe used radio resources.

Intra-MS" Hando*er

$n the intra/M#) handover when the >#) decides that handover is re0uiredbut the targeted cell is controlled by dierent >#) then it needs assistanceform the connected M#). $n comparison to the pervious handover discussedthe M#) mandatory for this kind of handover. ;esponsibilities of the M#) donot include processing the measurements of the >T# or M#) but to concludethe handover. This kind of handover can be other intra/M#) or $nter/M#). $nthe intra/M#) handover the targeted cell is allocate in dierent >#)

connected by the same M#). The M#) contacts the targeted >#) forallocation of the re0uired resources and inform the >#) when they are ready.After the successful resources allocation the M#) instructed to access thenew channel and the call is transferred to the new >#).

Inter-MS" Hando*er



The inter/M#) handover is performed when the two cells belonging todierent M#) in the same system. $n the inter/M#) handover the targetedcell is connected is connected to dierent M#) named as M#)/>! than theone currently serving the call M#) named as M#)/A!. 2igure shows G$$ theinter/M#) handover scenario.

:hen >#)/A determines that a handover is into another area its sends5N(F;4( to its connected M#)/A. 5N(F;4( contain information the )$ cell

identify! and the A) location area code! belonging to another M#) area.After identifying the correct neighbors M#) sends prepare handover to M#)/>. The J; of the M#)/> assign the temporary handover number and passesthe 5N(F;+4 to the target >#)/>. $n the response of 5N(F;+4 message>#)/> sends 5N(F;+4FA)K if the resources are available to M#)/>. The M#)/> forward to prepare for handover by sending >##A' message back to M#)A.$n the response M#)/A sends $AM which contains the handover numberbetween M#)/A and M#)/> and after getting the correspondent A)M fromM#)/>, M#)/A sends 5N(F)M( message to M#)/> to perform the handoverto the target >T#. The >#)/> sends 5N(F(+T that 5N(FA)) was receivedfrom the M#). The 5N(F(+T message is forwarded in a MA' to M#)/A. Now

the tra3c channel is established between M#)/A and M#)/>, and thetransport of the payload is carried to M#)/>, and the handover number isreleased by J; in M#)/>. >y receiving 5N(F)M' at M#)/> is signaled toM#)/A in a MA' message in #end +nd #ignal. This triggers M#)/A to send);F)M( message to >#)/A to release the radio resources

nam report

Documents