Location update for node movement between square regions at different levels The new location server region is in a square region, which is at a different level than the level of node A’
Trang 1Design and Analysis of a Multi-level Location Information Based
no of the node The previous level no is required by the new location server region in sending the new relative address of A, (i.e., current location server id and node id) to a location server region in the previous level This information is then relayed to all the other location server regions in the previous level Those location server regions after analyzing the current relative address of the node, find that the level no of node A has already changed, i.e., node A is no longer in the square region at their level Therefore, they delete the entry corresponding to node A from their database
Sub-region 3
Sub-region 1 Sub-region 2
Fig 11 Location update for node movement between square regions at different levels The new location server region is in a square region, which is at a different level than the level of node A’s previous square region Therefore, the new location server region must make a new entry in its location information database about the new fully qualified location information of node A This new location server region then needs to send the new relative location information of node A to other location server regions within the new square region These other location servers previously had no location information about node A Therefore, they need to make new entries in their location information database about the new relative location information of node A
4.2 Location query
Suppose node S wants to send a data packet to a destination node D but the location information of node D is unknown to S Corresponding to three location update scenarios three situations can evolve
Trang 2Mobile Ad-Hoc Networks: Applications
482
I Destination D is within same sub-region at same level as of source S:
In this case the location server region that is in-charge of the sub-region contains the fully qualified address of node D The source node S sends the data packet to the location server region The location server region extracts the current x and y coordinate position of node D from its fully qualified address and sends the data packet to node D at that location
Sub-region 0 Sub-region 3
Sub-region 1 Sub-region 2
S
D
Fig 12 Destination D is within same sub-region at same level as of source S
II Destination D is within other sub-region at the same level as of source S:
In this case the source S sends the data packet to the assigned location server region of its sub-region But as the destination D is within a different sub-region, therefore, the location server region of node S contains only the relative location information about destination D From this information, the location server region of node S can find the location server region, which is currently containing the fully qualified address of node D The location server region of node S then sends the data packet forwarded by S, to that particular location server region This new location server region ultimately sends the data packet to the destination node D
III Destination D is within other square region at different level than that of source S: The location server region now sends the data packet to the location server region of the square region that is encompassing the current level square region It also forwards the packet to the location server region of the square region that is contained by the current level square region The location server regions at other levels now follow the previously mentioned steps for location query This process is continued until the destination node D is found or the network boundary is reached Thus, if the destination node falls within the network boundary, the data packet is propagated from the source node S to the destination node D through the intermediate location server regions
Trang 3Design and Analysis of a Multi-level Location Information Based
Sub-region 1 Sub-region 2
S D
Fig 14 Destination D is within other square region at different level than that of source S
Trang 4Mobile Ad-Hoc Networks: Applications
484
5 Analysis of Layered Square Location Management (LSLM)
There are mainly two types of costs, which are important for any location management scheme These are - cost for location update and cost for location query When a node changes its position it must change its location information at the location server The number of packet forwarding operations it needs to perform per second, in order to maintain fresh location information, is known as the location updation cost Costupdate Similarly if a node wants to send a packet to a destination node whose location information
is unknown, in that case the sender node must perform location query, to find the location information of the destination node The number of packet forwarding operations that each node needs to perform for the purpose of location query defines the location query cost Costquery There is also a third type of cost, which is known as the storage cost The storage cost Coststorage signifies the number of location records that each of the location servers needs
to store
In the following sections we analyze these three types of costs for our proposed Layered Square Location Management (LSLM) scheme
5.1 Location updation cost [Cost update ]:
In our proposed scheme, location update has been divided into three parts As a consequence, the cost for location update can also be divided into three parts - i>Cost for location update for node movement within sub-region (Costupdate-intra-subregion) ii>Cost for location update for node movement between sub-regions (Costupdate-inter-subregion) iii>Cost for location update for node movement between square regions at different levels (Costupdate-inter-level)
Thus we can write,
Costupdate = Costupdate-intra-subregion + Costupdate-inter-subregion
Trang 5Design and Analysis of a Multi-level Location Information Based
The cost for location update depends upon the amount of forwarding load, where
forwarding load is determined by the number of hops traversed by a packet during location
update operation Thus the forwarding load, and as a consequence the cost will be greater
for a packet traveling a greater distance Cost for location update for node movement within
sub-region (Costupdate-intra-subregion) is basically the product of updation frequency and the cost
of updation of one location server region The cost of updation of one location server region
is proportional to the average number of hops an update packet takes to reach the assigned
location server region We denote this cost by Cost (1) We can approximate this cost by
considering the distance D=√2.2l.s; where l denotes level number (Fig 15)
Let us denote z as the average progress for each forwarding hop, where z is a function of the
radio transmission range rt and the node density (γ) (Seung-Chul.et al., 2001) We assume
both rt and γ are constants Therefore, z is also a constant It is possible to derive the average
number of hops an update packet takes by D/z If we consider the average velocity of a
node as v, and the transmission range of a node as rt,then the updation frequency is v/rt
If we assume S as the side length of the square region at the maximum level, i.e Lth level
square region, then, S ∞2L Thus, L ∞ log S Since, S ∞ √N, (N=Total Number of nodes in the
network), we have L ∞ log√N Thus,
Cost for location update for node movement between sub-regions (Costupdate-inter-subregion) is
the product of the boundary crossing rate (Ω) and the cost for updating the four location
server regions (Cost(4)) So,
Costupdate-inter-subregion = Ω.Cost (4)
The boundary-crossing rate is proved (Yu et al., 2004) to be proportional to v The cost of
updating four location server regions can be approximated by 4(Dl)/z Thus
Similarly we can formulate Costupdate-inter-level as
Costupdate-inter-level = Ω.Cost (8)
Trang 6Mobile Ad-Hoc Networks: Applications
Thus from “(1)”, “(2)” and “(3)” we have
Costupdate = Costupdate-intra-subregion + Costupdate-inter-subregion + Costupdate-inter-level = O (v.log√N)
5.2 Location query cost [Cost query ]:
If a source node has some data to send to a destination node, the source node must first
query a location server region to get the current location information of the destination
node The cost for this activity of querying the location information is known as location
query cost (Costquery) In order to calculate Costquery, we have to measure the expected
number of forwarding hops traveled by a query packet from the source node to its assigned
location server region, which can be approximated by D/z Therefore, the expected query
5.3 Storage cost [Cost storage ]:
In order to calculate the expected storage cost we need to find the average number of
records stored by a location server node in the network Dividing the total number of
records stored in the network by the total number of nodes acting as location servers gives
us the average number of records Each node in the network stores its address at the four
location server regions of its current layer of existence Earlier we have mentioned that each
location server region is a square area having side length of r Hence, the area covered by a
location server region can be expressed by r2 The average number of nodes (γ) is assumed
to be constant Thus the average number of nodes serving as location servers within a
location server region is r2 γ Now, the expected storage cost can be expressed as
Coststorage = (N.4 r2 γ)/(L 4 r2 γ) = N/L, where, N= Total number of nodes in the network; L= Maximum level number Since L ∞
log√N; the expected storage cost, Coststorage = O (N)
Trang 7Design and Analysis of a Multi-level Location Information Based
6 Conclusion
In this paper, we have presented Layered Square Location Management (LSLM), a novel scheme for the management of location information of the nodes in mobile ad hoc network The effectiveness of a location management scheme depends on reducing the costs associated with the major location management functions- location update and location query In case of a location service scheme we can reduce the location query cost by employing various caching strategies which is not possible for location update cost Keeping track of only the exact location information, makes location update highly expensive due to the high mobility of nodes In our scheme by dividing the entire network area into L levels
of square regions and using multi-level location information, we have been able to provide a unique way to reduce the cost associated with both location update and location query Further investigation on performance analysis of this scheme in different network scenarios can be taken as extended work
7 References
Amouris, K.N.; Papavassiliou, S & Li, M (1999) A position-based multi-zone routing
protocol for wide area mobile ad-hoc networks In Proceedings of the IEEE Vehicular
Technology Conference (VTC), pages 1365-1369
Chen, T W & Gerla, M (June 1998) Global State Routing: A New Routing Scheme for Ad
Hoc Wireless Networks, Proceedings of IEEE ICC 1998, pp 171-175
Cheng, Christine T.; Lemberg, H L.; Philip, Sumesh J.; Berg, E van den & Zhang, T
(March 2002) SLALoM: A scalable location management scheme for large mobile
ad-hoc networks In Proceedings of IEEE WCNC
Chiang, C C.; Wu, H K.; Liu, W & Gerla, M (April 1997) Routing in Clustered Multi-Hop
Mobile Wireless Networks with Fading Channel, Proceedings of IEEE SICON 1997,
pp 197-211
Clausen, T H.; Hansen, G.; Christensen, L & Behrmann, G (September 2001) The
Optimized Link State Routing Protocol, Evaluation Through Experiments and
Simulation, Proceedings of IEEE Symposium on Wireless Personal Mobile
Communications 2001
Dube, R.; Rais, C D.; Wang, K Y & Tripathi, S K (February 1997) Signal Stability-Based
Adaptive Routing for Ad Hoc Mobile Networks, IEEE Personal Communications
Magazine, pp 36-45
Garcia-Luna-Aceves, J J & Spohn, M (October 1999) Source-Tree Routing in Wireless
Networks, Proceedings of IEEE ICNP 1999, pp 273-282
Gerla, M.; Pei, G & Hong, X (August 2000) Lanmar: Landmark routing for large scale
wireless ad hoc networks with group mobility In Proceedings of the First IEEE/ACM
Workshop on Mobile Ad Hoc Networking and Computing (MobiHOC)
Haas, Z J (October 1997) The Routing Algorithm for the Reconfigurable Wireless
Networks, Proceedings of ICUPC 1997, vol 2, pp 562-566
Haas, Z J & Pearlman, M R (August 1998) “The zone routing protocol (ZRP) for ad hoc
networks (Internet Draft)”
Iwata, A.; Chiang, C C.; Pei, G.; Gerla, M & Chen, T W (August 1999) Scalable Routing
Strategies for Ad Hoc Wireless Networks, IEEE Journal on Selected Areas in
Communications, vol 17, no 8, pp 1369-1379
Trang 8Mobile Ad-Hoc Networks: Applications
488
Joa-Ng, M & Lu, I T (August 1999) A Peer-to-Peer Zone-Based Two-Level Link State
Routing for Mobile Ad Hoc Networks, IEEE Journal on Selected Areas in
Communications, vol 17, no 8, pp 1415- 1425
Johnson, David B & Maltz, David A (1996) ( Dynamic source routing in ad hoc wireless
networks In Imielinski and Korth, editors, Mobile Computing, volume 353 Kluwer
Academic Publishers
Li, J.; Jannotti, J.; Couto, D De.; Karger, D & Morris, R (August 2000) A scalable location
service for geographic ad-hoc routing In Proceedings of ACM MobiCom, pages
120.130
Murthy, S & Garcia-Luna-Aceves, J.J (October 1996) An efficient Routing Protocol for
Wireless Networks ACM Mobile Networks and App Journal, Special Issue on Routing
in Mobile Communication Networks
Park, V.D & Corson, M.S (April 1997) A Highly Adaptive Distributed Routing Algorithm
for Mobile Wireless Networks Proceedings of IEEE INFOCOM’97, Kobe, Japan
Perkins, Charles E & Bhagwat, Pravin (August 1994) Highly dynamic
Destination-Sequenced Distance-Vector routing (DSDV) for mobile computers In Proceedings of
the SIGCOMM ’94 Conference on Communications Architectures, Protocols and Applications, pages 234–244
Perkins, Charles & Royer, Elizabeth (1999) Ad-hoc on-demand distance vector routing In
Proceedings of IEEE Workshop on Mobile Computing Systems and Applications
Seung-Chul.; Woo, M & Singh, Suresh (2001) Scalable routing protocol for ad hoc
networks Wireless Networks, 7(5):513-529
Sinha, P.; Sivakumar, R & Bharghavan, V (August 1999) CEDAR: A Core Extraction
Distributed Ad Hoc Routing Algorithm, IEEE Journal on Selected Areas in
Communications, vol 17, no 8, pp 1454-1466
Sisodia, R S.; Manoj, B S & Murthy, C Siva Ram (March 2002) A Preferred Link-Based
Routing Protocol for Ad Hoc Wireless Networks, Journal of Communications and
Networks, vol 4, no 1, pp 14-21
Su, W & Gerla, M (December 1999) IPv6 Flow Handoff in Ad Hoc Wireless Networks
Us-ing Mobility Prediction, ProceedUs-ings of IEEE GLOBECOM 1999, pp 271-275
Toh, C K (March 1997) Associativity-Based Routing for Ad Hoc Mobile Networks, Wireless
Personal Communications, vol 4, no 2, pp 1-36
Xue, Y.; Li, B & Nahrstedt, K (2001) A scalable location management scheme in mobile
ad-hoc networks In Proceedings of the IEEE Conference on Local Computer Networks (LCN
'01)
Yu, Yinzhe.; Lu, Guor-Huar & Zhang, Zhi-Li (2004) “Enhancing location service scalability
with HIGH-GRADE,” Dept of Comp Sci & Eng., University of Minnesota, Technical Report TR-04-002
Trang 9Power Control in Ad Hoc Networks
Muhammad Mazhar Abbas and Hasan Mahmood
to ad hoc networks has many challenges and implementation complexities (Chauh & Zhang,2006) (Basagni et al., 2004) The power control is of great significance in ad hoc networks
implementation of effective power control techniques, the ad hoc network can improve theirvital parameters, such as power consumption, interference distribution, throughput, routing,connectivity, clustering, backbone management, and organization (Basagni et al., 2004)
We discuss several power control algorithms commonly used in ad hoc networks to get insight
of power control techniques and their effectiveness Most of the algorithms are adapted fromcellular networks, modified accordingly, and proposed for ad hoc networks Moreover, weargue the enhancement in performance of ad hoc networks with the use of these power controlalgorithms
The power control requirements vary depending on the physical and network layer
prevailing power control algorithms to different physical layer models and discuss theirperformance The application to CDMA based networks is emphasized as these types ofnetworks have strict power control requirements and the performance is severely degradedwithout appropriate power control In cellular networks, the power control requirements arestringent, especially in multiple access technologies The appropriate allocation of power tothe transmitters facilitates interference control and saves energy
The near-far effect starts to dominate as the transmission power levels are not properlymanaged The advantage of cellular networks over ad hoc networks is the presence of centralmanagement, and as a consequence, the uplink power control can be achieved This is incontrast to ad hoc networks, which lack central management and most of the nodes are inpeer to peer configuration (Blogh & Hanzo, 2002)
In addition, transmit power control is a cross layer design problem affecting all layers ofthe OSI model from physical layer to transport layer (Jia et al., 2005) In general, powerconservative protocols are divided into two main categories: transmitter power controlprotocols and power management algorithms Second class can be further divided into MAClayer protocols and network layer protocols (Ilyas, 2003)
22
Trang 102 Theory and Applications of Ad Hoc Networks
At the end of the chapter, we discuss the concept of joint power control and routing
in ad hoc networks Power can be controlled in ad hoc networks by choosing optimalroutes The existing routing protocols may be classified as, uniform, non-uniform, proactive,
reactive, hybrid, source, and non-source routing protocols (?Chaudhuri & Johnson, 2002) To
further explain joint power control and routing techniques, we discuss a Minimum AverageTransmission Power Routing (MATPR) technique (Cai et al., 2002), which implements apower control routing protocol using the concept of blind multi-user detection to achievethe task of minimum power consumption The Power Aware Routing Optimization (PARO)technique (Gomez et al., 2003), a protocol for the minimization of transmission power in adhoc networks, is based on the concept of node to node power conservation using intermediatenodes, usually called redirectors PARO is efficient in both static and dynamic environmentsand is based on three main operations: overhearing, redirecting, and route maintenance
2 Cellular networks
The wireless cellular networks require a fixed and well defined infrastructure This type ofnetwork infrastructure is suitable to efficiently manage the network operations Generallythe network can be managed and operated by a central operations point In the field, thethe physical parameters, such as transmission frequency, resource allocation, and powercontrol parameters are monitored and controlled by base station which have fixed location
We focus on power control for these types of configurations in order to study and analyzeimplementation to ad hoc networks
Power control is a necessary feature in cellular communication networks with multiple accesstechnologies Power control has many management features such as interference control,energy saving, and connectivity (Almgren et al., 2009) In power control mechanism eachuser transmits and receives at an appropriate energy level, i.e., the transmission powers arecontrolled in such a way that the interference is minimized, while achieving sufficient quality
is created as the signals of mobile propagate through different channels before reaching theircorresponding base station (Moradi et al., 2006) The purpose of power control is to allow allmobile signals to be received with same power at the base station Uplink power controlenhances capacity of networks (Gilhousen et al., 1991) On the other hand, in downlinktransmission, the near-far effect problem is not as important, because signals from the basestation reach the mobile station while propagating through same channel (Lee et al., 1995).Uplink power control algorithms achieve their functions through open loop and closed looppower control, which can be further divided into closed outer loop power control and closedinner loop power control In open loop power control, the mobile user adjusts its transmissionpower based on the received signaling power from the base station (Chockalingam & Milstein,1998) In closed-loop power control, based on the measurement of the link quality, the base
490 Mobile Ad-Hoc Networks: Applications
Trang 11Power Control in Ad Hoc Networks 3
station sends a power control command instructing the mobile to increase or decrease its
transmission power level and sets the target signal-to-interference ratio (SIR) to such a level
that sufficient quality of service is guaranteed (Rintam¨aki, 2002)
Power can be controlled in a centralized or distributed fashion In centralized form a controllermanages the information of all the established connections and channel gains, and controlsthe transmission power level (Grandh et al., 1993) While in the distributed form a controllercontrols only one transmitter of a single connection It controls transmission power based
on local information such as the signal to interference ratio and channel gains of the specificconnection Distributed form of power control is easy to use in common practice because itdoes not require extensive computational work (Zender, 1993)
Although we aim to discuss power control techniques for wireless ad hoc networks, it
techniques were initially applied to cellular networks, and with the advent of ad hoc networkwere adapted and modified to meet new requirements Some of the basic power controlalgorithms are presented below which are related to wireless cellular networks and theirimplementations
2.1 Power control as eigen value problem
In the era of 1980s the concept of Signal to Interference Ratio (SIR) balancing in power
control algorithms for cellular networks based on Code Division Multiple Access (CDMA)and other technologies were used by researchers (Nettleton, 1980) (Nettleton & Alavi, 1983)(Alavi & Nettleton, 1982) Initially, the power control problem was focused and treated as an
eigen value problem with a non negative matrix G and corresponding balance power vectors
pu and p dwhich satisfy the eigen value problem as
and
whereγuandγ d are desired uplink and downlink SIRs By taking λ(G)as eigen value of G a
solution to the above problem is given as
Another solution to SIR balancing problem is given as
where spectral radiusρ is such that ρ>1
(Foschini & Miljanic, 1993) to solve the above eigen value problem iteratively is by solving
liner algebraic equations, represented as AP=b, where P= [p1, p2, p N]T, and
Trang 124 Theory and Applications of Ad Hoc Networks
A generalized frame work for convergence is given in (Yates, 1995) By using proper powercontrol, the interference is eliminated and we get iteration as
p i(k+1) =γ tar
where p i is the power of i thuser andγ tar
i is the target SIR
2.2 Distributed power control techniques
The Distributed Power Control (DPC) algorithm is applied at individual nodes in the networkand the objective is to converge system power allocations to a suitable level (Grandhi et al.,1994) This can be accomplished by using feedback power control (Ariyavisitakul, 1994) Inthis method the power is adjusted in steps which may have fixed or variable size It is seenthat the performance of a power control algorithm with fixed step size and variable step size
is almost the same In addition, the higher power control rate can accommodate the effect offast fading
With the implementation of distributed power control, the SIR of the system can be controlled
and managed to some extent As a result, the outage probability of an individual link or a set
of links can be reduced or entirely eliminated The implementation of this type of methodrequires a distributed power control algorithm which reduces the outage probability to zero
by keeping SIR above threshold value (Zander, 1992).
In another approach, a smaller balancing systems can be constructed by turning thetransmitter of cells off so the outage probability is minimized In some scenarios, if the value
of SIR for a mobile is less than threshold value then outage probability is reduced and mobile
is dropped from network (Wu, 1999) This improves the remaining network SIR.
An optimal SIR based distributed power control technique can be used by unconstrained and
constrained optimization (Qian & Gajic, 2003) The theme of this algorithm is to establish a
proportionality between transmission power and the error between the actual SIR and the desired SIR Difference of transmission power from time step k to k+1 is given as
2.3 Discrete time dynamic optimal power control
In this method, the reverse link system information is used for power control A cost function,consisting of weighted sum of powers and some additional parameters is defined An optimalpower control law is presented based on a cost function comprising of weighted sum of power,
power update information, and SIR error It is also assumed that there is no significant change
492 Mobile Ad-Hoc Networks: Applications
Trang 13Power Control in Ad Hoc Networks 5
in SIR from one step to the next For this purpose, a technique named as discrete time dynamic
optical control is implemented (Koskie & Gajic, 2003) The general cost function and sufficientconditions for optimality are defined as
where J is the controller, L is the cost function and H is the hamiltonian Some of the different
optimal controllers for three cost functions are
2.4 Linear and bilinear power control techniques
The optimization of power conservation results in improved SIR distribution for the entire
network Although these optimizations are based on some estimates, as a consequence, errorsare introduced in the actual results (Gajic et al., 2004)
The power control techniques named as linear and additive power updates algorithm and
bilinear control algorithm are based on optimization of SIR error It can be seen that mobile
power is updated by using a distributive linear control law, given as
where i=1, 2, n By minimizing SIR error and after other calculations the optimized power
updates can be obtained as
Trang 146 Theory and Applications of Ad Hoc Networks
2.5 Power control technique based on relaxation method
This method is particularly useful in networks with multiple access technology, such asCDMA A relaxation method can be used in solving iterative power control techniques Twocommon techniques for iterative solution of power control problems can be used effectivelywith relaxation method Application to Jacobi iteration method and Gauss Siddel iterationmethod for solution of power control problem, by introducing a relaxation parameter in thesetechniques, is presented as a modified Jacobi iteration
technique for solution of power control problem The algorithms implemented by relaxationmethod converge faster than simple distributed power control algorithm (Siddiqua et al.,2007)
2.6 Distance based power control technique
The distance between transmitters and receivers can be estimated in a wireless networks Theattenuation of the signals is proportional to the distance which they travel Therefore, if theinformation about the distances is know in real time or a prior, the power can be adjustedefficiently (Nuaymi et al., 2001) If a base station is present, the transmit power of each mobilestation can be controlled by using distance information between base station and mobile
stations This algorithm computes the transmitted power P m of a mobile node m as
2.7 Kalman filter based power control technique
In an uplink closed loop power control algorithm based on Kalman filter technique, the
controller or a base station estimates SIR in a closed loop system (Rohi et al., 2007) The SIR
can be estimated by any suitable method The outage probability calculated by this method
is smaller as compared to others According to algorithm details, the base station estimates
the SIR for a user and provide as input to Kalman predictor Its output is compared with the desired SIR and the difference is quantized by a PCM The transmitted power of user is then updated and SIR estimation is given as
494 Mobile Ad-Hoc Networks: Applications
Trang 15Power Control in Ad Hoc Networks 7
and
The outage probability is given as P0=P r(SIR r<SIR0) Where SIR r is measured at base
station and SIR0is the minimum value of SIR for achieving desired BER.
2.8 Power control technique based on linear quadratic control theory
The state-space formulation and linear quadratic control technique can be used to solve theproblem of power control by considering each mobile to base station link as an independentsubsystem described as
The input to each subsystem U i(n)depends on the total interference produced by other users
plus the noise in the system and each S i(n)track is made equal to the threshold value of SIR
(Osery & Abdallah, 2000) For the discrete case the new state is given by
The feedback controller V i(n) = −[k ς k s]x i(n) +k s γ∗, where[k ς k s]is the gain matrix whichare found by solving the Riccati equation If the right feedback gains [kςks]is chosen, the
steady-state S i(n)will go to the threshold SIR To find the optimum feedback control for the
state-space representation given above, the Linear Quadratic Control theory is used After thegain matrix[k ς ks]is found, the power control can be expressed as
Pi(n+1) =min[Pi , S i(n+1)Ii(n)] (32)
The method assures that the maximum transmission power of the mobile i will not be
exceeded This method reaches a zero outage probability with less iterations than otherdistributed power control methods This approach was also found to be more effective inhandling a large number of mobile stations in the system
495
Power Control in Ad Hoc Networks
Trang 168 Theory and Applications of Ad Hoc Networks
2.9 Power control technique based on utility and pricing
The power control algorithm can be implemented in a distributed fashion based on utilityand pricing concepts (Shah et al., 1998) The efficiency of this protocol can be improved in low
BER and high SIR conditions The formula of SIR of user j at base station k is given as
In this method, by introducing a pricing factor the utility is maximized and as a result helps
in power control problem A general utility function which is a monotonically increasing
function of SIR is given as
u j= E
Where f(γ)is a measure of efficiency of protocol The power control problem is considered
as a cooperative power control game The user maximizes its utility at equilibrium point
with maximum SIR value as Max u i(p1, p2, p N),∀i=1, 2, N, and f(γ∗) =γ∗f(γ∗)
We can also consider a monotonically increasing pricing function, F=βp j, which is assumed
to depend upon a cost function, given as,
for both voice and data users The value of SIR for user i with transmission power P can be
written as
SIR i= G ii P i
The main goal of this algorithm is to maximize the net utility by transmission power
adjustment and softening the hard SIR requirements as
NU i(SIR i , P i) =U i(SIR i) −C i(P i) (40)
where C i(P i) =α i P i is assumed cost function of power for the user i The power control
user i is
496 Mobile Ad-Hoc Networks: Applications
Trang 17Power Control in Ad Hoc Networks 9
Thus, by using utility based power control protocol, a user can control its power by decreasing
its SIR and even turn off transmission during heavily loaded network.
2.10 Opportunistic power control technique
In this distributed opportunistic power control algorithm, the transmission power depends
on channel gain by observing feedback from the receiver The transmission rate is managed
by SIR at the receiver (Leung & Sung, 2006) The SIR of a terminal i in a cellular system
R i , where R i is the effective
interference to terminal i, and is given as
(45)
This algorithm converges and equation P i n R n i =ς i is satisfied The transmission power of
terminal i varies directly with ς i
2.11 Power control technique based on simple prediction Method
A simple prediction is sometimes useful for power control in wireless networks (Neto et al.,2004) This approach can be used to implement a distributed power control algorithm, based
on simple prediction method, and by considering both path gain and SIR as time varying functions using Taylor series Discrete-time SI NR is given as
γ i(k) =g i(k)p i(k)
where p i(k) =I i (k)γ i (k)
g i (k) is known as necessary transmission power The transmission power at
instant k+1 is given by using Taylor series as