A REVIEW ON VARIOUS NETWORK RESULTS FOR REPORTING CELL PLANNING IN GENETIC ALGORITHM TECHNIQUE

(1)

A REVIEW ON VARIOUS NETWORK

RESULTS FOR REPORTING CELL

PLANNING IN GENETIC ALGORITHM

TECHNIQUE

C.Ashok Baburaj, M.C.A, M.Phil, (P.hD)

Assistant Professor, Dept of MCA, KLN College of Engineering, Sivagangai District, Dr. K. Alagarsamy,

Associate Professor, Computer Centre, Madurai Kamaraj University, Madurai

ABSTRACT :

The Location Management problem is an important issue in mobility management. One of the most common strategies of location management is to designate some of the cells in the network as “reporting cells” and the other cells as “non-reporting cells”. Mobile terminals update information about its current location in a database when it enters into a new reporting cell. In this paper a new concept has been introduced to minimize the location management cost by maintaining the mobility history. Genetic Algorithms are stochastic methods that can be used to solve a very broad class of optimization problems. A Genetic Algorithm technique has been used to solve the various network results which shows the effectiveness of the proposed approach for the Reporting Cell Planning in Mobile Computing.

Keywords : Reporting Cell Planning; Location Management; Genetic Algorithm

1. Introduction

(2)

of the reporting cell that was last reported by the mobile terminal. The key issue of the reporting cell planning scheme is how to select a set of reporting cells to minimize the total location management cost.

2 . Location Management Cost

To determine average cost of a location management in reporting cell planning strategy, one can associate a cost component to each location update performed, as well as to each polling/paging of a cell. The total cost of the above two cost components(location update and cell paging) over a period of time T, can then be averaged to give the average cost of a location management strategy.[12] Therefore the formulae used to calculate the total cost of a location management approach is

Total Cost = C.N

LU

+ N

P

Where NLU denotes the number of location updates performed during time T. NP denotes the number of paging

performed during time T, and C is a constant representing the cost ratio of location update and paging .It is recognized that the cost of a location update is usually much higher than the cost of paging – several times higher [7]. In light of this fact, we use C = 10 [5] [14].

Most of today’s wireless network consists of cells. Each cell contains a base station, which is wired to a fixed wire network. These cells are usually represented as hexagonal cells resulting in six possible neighbors for each cell. In reporting cell approach, a subset of base stations called reporting centers is selected among all base stations. The cells associated with these base stations are referred to as reporting cells. The reporting centers periodically transmit short messages on the wireless channel to identify their roles. Thus, a mobile terminal can find out whether or not it is in a reporting cell by listening to the transmitted message.

The vicinity of reporting cell i is the collection of all the cells that are reachable from a reporting cell i without crossing another reporting cell . When a call arrives for a mobile terminal, the search is restricted within the vicinity of reporting cell to which mobile terminal last reported.

3

1

6

4

2 5 7

Figure : 1 7 cells network with reporting cells.

An example in Figure 1 shows a network configuration in which, cells 2, 3, 4, and 7 are chosen as reporting cells (shown as shaded cell) [3] . Cell 1, 5, and 6 are reporting cells. Cell 2 is directly connected to non-reporting cells 1 and 5, and there are no non-reporting cells between them. Thus, the vicinities of cell 2 are cells 1, 5, and 2. For the same reason, the vicinities of cell 3 are cells 1, 6 and 3; the vicinities of cell 7 are cells 5, 6 and 7; the vicinities of cell 4 are cells 1, 5, 6 and 4. Each non-reporting cell can also be assigned a vicinity value. However, it is clear that a non-reporting cell may belong to the vicinity of several reporting cells, which may have different vicinity values. For example, in Figure 1, cell 1 belongs to the vicinity of reporting cells 2, 3, and 4, with vicinity values 3, 3, and 4 respectively. For location management cost evaluation purposes, the maximum vicinity value will be used. As such, in this case, the vicinity value of 4 is assigned to cell 1.

In Reporting Cell planning, with each cell i a movement weight, and call arrival weight, denoted wmi and wci, respectively are associated. The movement weight represents the frequency or total number of movements into a cell, while the call arrival weight represents the frequency or total number of call arrivals within a cell. Formulas for the total number of location updates and paging (performed during a time period T) are given as follows:

NLU = ∑ wmi ---(1) iєS

N-1

(3)

Where wmi denotes the movement weight associated with cell i, wcj denotes the call arrival weight associated with cell j, v(j) denotes the vicinity value of cell j, N denotes the total number of cells in the network, and S denotes the set of reporting cells in the network. Using equation (1) and (2), the formula to calculate the location management cost of a particular reporting cells configuration is given as follows:

N-1

Total Cost =C* ∑ w mi + ∑ wcj . v(j) ---(3) iє_{S J=0}

Where C is a constant representing the cost ratio of location update and paging.

3 The Proposed System

In the proposed scheme a new technique is being introduced , which maintains the history or mobility pattern (of size h) of the last visited reporting cells [1]. The updating does not take place when the user roams within the reporting cells of his mobility pattern. That is the location information is updated when the user enters to a new reporting cell, which is not in his history. As a result, the updation cost is proportionately reduced with the value of h (no. of entries in the history). When we increase the number of reporting cells in the history, the location update cost is proportionately reduced.

Hence the cost equation can be modified as follows NLU = ∑ NWmi

iєS Where NWmi is the new movement weight.

NWmi = Wmi * (S-h)

(S-1)

Where h is the number of reporting cells maintained in the history Here if we keep h=1, the NWmi tends to NWmi By increasing h value the NWmi will be reduced, as a result the updating cost is reduced.

Consequently, the paging cost gets increased proportionately to the h value. But it can be kept under control by using the following technique. Whenever the user enters into the reporting cells of his previous history, the mobility pattern in the mobile is modified and does not leads to the location update.

Whenever a call arrives to the user, the user may be available within the vicinity of any one of the reporting cells in the history. This increases the number of cells to be searched. But it is only for the first time call, because after searching the user among the list of cells that can be reached from the reporting cells in the mobility pattern, the new mobility pattern which is maintained in the mobile is updated to the server. So the next call to the user doesn’t take that much number of searches.

The new paging cost equation is obtained from the new call arrival weight. NWcj =Wcj * NWmi

Wmi

Search Cost for the Location updated users:

N-1

NP1= ∑ (NWcj) * v(j) ---(4)

j=0

Search Cost for the non-updated users from the same reporting cell: N-1

(4)

Search Cost for non-updated users from different reporting cell: (first call)

N-1

NP3 = ∑ (Wcj - NWcj)*(S-1)/S*v(j)*h/2 ---(6)

j=0 Call factor

(Subsequent calls)

N-1

NP4 = ∑ (Wcj - NWcj)* (S-1)/S *

j=0

(1-1/Call factor) * v(j) ---(7)

The call factor can be calculated as follows, if (Wcj / Wmi)< 1 then call factor=1 else call factor = (Wcj / Wmi ).

The Total Paging Cost will be

Np’ = Np1 + Np2 + Np3+ Np4

Total Cost = C.NLU’ + NP’

The totals paging cost is divided into four sub components and except the third component (Np3) the other component costs are similar to the old method. As a result the increase in the paging cost is under control; hence it improves the total cost for the reporting cell configuration. The cost function described above shows that by varying the size of the mobility pattern (h), the total cost can be reduced to some extent. If h=1 then this cost is equivalent to the old cost. So in the worst case it behaves like the old scheme and in best case, we can introduce h value so that the entire cost is reduced.

The problem of marking the cells in the network as reporting cells and deciding the value of h can be seen and solved using one of the artificial life techniques like GENETIC ALGORITHM.

4. Genetic Algorithm Technique for Reporting Cell Planning Problem

A Genetic Algorithm(GA)[4] [8][9] [13]is a biologically inspired optimization and search technique developed by Holland [6]. A genetic algorithm is an iterative procedure that consists of a constant-size population of individuals, each one represented by a finite string of symbols, known as the genome, encoding a possible solution in a given problem space. This space, referred to as the search space, comprises all possible solutions to the problem at hand. The genetic algorithm is applied to spaces which are too large to be exhaustively searched.

(5)

The execution steps in Genetic Program are as follows :

1. Randomly create an initial population (generation 0) of individual computer programs composed of the available functions and terminals.

2. Iteratively perform the following sub –steps (called a generation) on the population until the termination criteria is satisfied

a) Execute each program in the population and ascertain its fitness (explicitly or implicitly)using the problem’s fitness measure

b) Select one or two individual program(s) from the population with the probability based on fitness (with reselection allowed) to participate in the genetic operation in (c)

c) Create new individual program (s) for the population by applying following genetic operations with specified probabilities

(i) Reproduction – copy the selected individual program to the new population.

(ii) Crossover - create new offspring program (s) for the new population by recombining randomly chosen parts from two selected programs.

(iii) Mutation – create one new offspring program for the new population by randomly mutating a randomly chosen part of one selected program

3) After the termination criteria is satisfied , the single best program in the population produced during the run(the best -so-far individual) is found and designated as the result of the run . If the run is successful , the result will be a solution to the problem

4.1 Pseudo-code of the Genetic Algorithm for the Reporting Cell Planning

begin GA

g:=0 { generation counter } Initialize population P(g) while not done do

Evaluate population P(g) { i.e., compute fill vicinity } Compute Objective function { i.e., cost function ) Compute Expected Count

Reproduce P(g) Crossover P(g) Mutate P(g) g:=g+1; end while

end GA

5.1 REVIEW OF 4 X 4 NETWORK RESUTS

INPUT DATA SETS FOR 4 X 4 NETWORK OUTPUT DATA SETS FOR 4 X 4 NETWORK CELL WCI WMI

0 517 518 1 573 774 2 155 153

4 642 1617

5 951 472 6 526 650

9 224 2149

11 600 952 12 25 307 14 695 1346

(6)

Inference : In 4 x 4 Network

The Maximum Number of cells = 16 The Number of Generations = 100 The Minimum Value = 7.203 The Average Value = 9.401 The Maximum Value = 10.470

5.2 REVIEW OF 6 X 6 NETWORK RESUTS

INPUT DATA SETS FOR 6 X 6 NETWORK OUTPUT DATA SETS FOR 6 X 6 NETWORK

CELL WCI WMI

0 714 1039

1 120 1476

2 414 262

3 639 442

4 419 1052

9 221 296

14 789 1479 19 682 1368 24 328 16 29 769 747

HISTORY DATA SETS

1 10.662 2 10.410 3 10.789 4 10.790 5 10.673 10 10.662 15 10.662 20 10.711 25 10.662 30 10.797

Inference : 6 x 6 Network

GRAPH FOR 4 X 4 NETWORK DATASETS

0 5 10 15 20

1 2 3 4 5 6 7 8 9 10

D A T A S ET S

HISTORY DATA SETS

GRAPH FOR 6X6 NETWORK DATASETS

0 10 20 30 40

1 2 3 4 5 6 7 8 9 10

DATA SETS

HIS

T

O

R

Y

(7)

5.3 REVIEW OF 8 X 8 NETWORK RESUTS

INPUT DATA SETS FOR 8 X 8 NETWORK OUTPUT DATA SETS FOR 8 X 8 NETWORK

CELL WCI WMI

0 968 533

4 902 1336

9 212 1071

19 25 1256 24 694 655 29 96 1081 34 132 1011 44 362 672 49 135 1400 59 896 1017

HISTORY DATA SETS

1 11.603 5 12.080 10 11.875 20 12.241 25 11.948 30 11.828 35 11.754 45 11.686 50 12.022 60 11.798

Inference : In 8 x 8 Network

6 . Conclusion

This research paper analysis the various network results for reporting cell planning, It reviews the result for 4x4, 6x6 and 8x8 Network. The solution to the reporting cell planning problem can be improved by using the user mobility pattern and this concept is proved by implementing through Genetic Algorithm. A Genetic Algorithm technique has been used to solve the various network results which shows the effectiveness of the proposed approach for the Reporting Cell Planning in mobile computing.

References

[1] C.Ashok Baburaj and S.Thabasu Kannan “Effective and Efficient Mining of Data in Mobile Computing” – IAENG International Journal Of Computer Science , 32:4 IJCS_32_4_5 , 2006

[2] N.A.Bar and I.Kessler, “Tracking Mobile Users in Wireless Communications Networks”, IEEE Trans. Information Theory, Vol. 39, PP. 1877-1886, 1993.

[3] Falguni Mehta, Prashant Swadas :” A Simulated Annealing Approach to Reporting Cell Planning Problem of Mobile Location Management”, International Journal of Recent Trends in Engineering, Vol 2, No. 2, November 2009

[4] D.E.Goldberg, Genetic Algorithms in Search, Optimization, and Machine Learning Mass: Addison – Wesley, 1989.

[5] R.L. Gondim, “Genetic Algorithms and the location area partitioning problem in cellular networks”, proc. IEEE 46th vehicular technology conf., 1996.

[6] J.H.Holland, Adaptation in Natural and Artificial Systems . Ann Arbor:Univ. of Michigan Press, 1975

[7] Imielinski and B.R. Badrinath, “Querying Locations in Wireless Environments”, Proc. Wireless Comm. Future Directions, 1992. [8] Z.Michalewicz, Genetic Algorithms + Data Structures = Evolution Programs. Berlin : Springer – Verlag, 1994.

[9] M.Mitchell, An Introduction to Genetic Algorithms . Cambridge, Mass .: MIT Press, 1996

[10] D.Plassmann, “Location Management Strategies for Mobile Cellular Networks of 3rd Generation “, Proc. IEEE 44TH Vehicular Technology Conf., 1994.

[11] R. Subrara & A.Y. Zomaya, ” A comparison of three artificial life techniques for reporting cell planning in mobile computing”,IEEE trans. On parallel and distributed systems, vol 14, no. 2, Feb 2003.

[12] R. Subrara & A.Y. Zomaya, “Location Management in mobile computing”, proc. ACI/IEEE International conf. computer systems and applications, 2001.

[13] M.D.Vose, The Simple Genetic Algorithm : Foundations and Theory. Cambridge , Mass.: MIT Press, 1999

[14] H.Xie, S.Tabbane, and D.J.Goodman, “Dynamic Location Area Management and Performance Analysis”, Proc. 43rd IEEE Vehicular Technology Conference Personal Comm. Freedom Through Wireless Technology, 1993

[15] K.L.Yeung and T.S.P. Yum, “A Comparative Study on Location Tracking Strategies in Cellular Mobile Radio Systems”, Proc. IEEE Global

GRAPH FOR 8X8 NETWORK DATA SET

0 10 20 30 40 50 60 70

1 2 3 4 5 6 7 8 9 10

D A T A SET S

HISTORY