Proposal of an opportunistic approach for managing Mobility in Wireless community networks based on the Markov model

Today, there has been considerable progress in network and mobile radio technology. We are witnessing the deployment of different standards of mobile networks such as GSM, UMTS and wireless, such as WI-FI (IEEE802.11) and WIMAX (IEEE 802.16). Wireless community networks have grown rapidly with the emergence of Wi-Fi thanks to their simplicity, speed and low deployment cost. They are therefore an interesting alternative to conventional local networks. The management of mobility in wireless networks is now a challenge in this type of networks with the proliferation of mobile terminals. A mobile user can change his or her network access point by moving, without interrupting the current service session, that is, undergoing handover. Our study is located at the top layer of the OSI model (network, application and transport) from where we choose the SCTP / mSCTP protocols of the transport layer to support mobility thanks to multihoming with dynamic address reconfiguration RDA), the control of the borrowed addresses (or paths), the transfer of data in an association. And negligible losses of packets. We have modeled the mobility management using the Markov chain. Through the studies, on the functions of aggregations we also revealed the transmission rates which shows a better quality of service (QoS).

This community-based wireless network architecture includes: Wireless (WI-FI) access on all sites; The external access points Mesh 802.11a / b / g level 2 2.4GHz / 5GHz; Hot Spots for WiFi Internet access; The SSL VPN ensures security of access to internal network resources by using an SSL tunnel between the client (connected to a public network) and the VPN box. This paper makes a choice on a mobility protocol that meets our context of managing mobility in wireless community networks. After studying the high layers of the OSI model (network, application and transport), the choice is made on the transport-level mSCTP protocol for advantages such as multihoming over network-level protocols (MIP) and application (SIP). The Markov chains to model the mSCTP protocol and the aggregation functions to show the audio quality of service (QoS) is used. The next part of this article is presented as follow: the approach that is used to model the mSCTP mobility management protocol is described in part 2, and the results are shown in part 3. before concluding the work in part 5, a brief summary has to be done in part 4.
2. Methodology 2.1 Protocols of mobility management: State of the art In this section we present a synthesis of the most relevant mobility protocols in the upper layers of the OSI model: network, transport and application. the advantages and disadvantages of each protocol and justify our choice of the mSCTP protocol is presented. 2.1.1 Mobility at network level: Protocol Mobile IP (MIP) Mobile IP (MIP) [3,4] is presented as the current solution to the breakdown in communication problems during the movements of mobile nodes in IP networks. This protocol allows mobile nodes to move from networks to networks without breaking their ongoing sessions. The mobile node gets a new temporary IP address to each entry in a visited network, thanks to DHCP protocol and registers with its mother agent and the agent visited of the new network. This address indicates the current position of the mobile node. It will communicate with its home agent, which undertakes to intercept packets in the core network of the mobile node and to transmit it to its current position. Mobile IP can be divided into Mobile IPv4 (MIPv4) and Mobile IPv6 (MIPv6) depending on the version of the associated IP. Both protocols basically provide similar functionality with a few exceptions in the retail of operating mechanisms. Although the Mobile IPv6 protocol allows to solve the triangular routing problem packet used in the Mobile IPv4 protocol, it still suffers from several weaknesses. These weaknesses, are: The time of handover is long. Specifically, the period of motion detection phase, the phase of the auto-configuration of addresses and of the association update phase are very long for real-time applications; the loss of Packet during the handover can be significant [5]. ( • In the case of retransmission of chunk towards a multihomed endpoint, the receiver must choose a destination address other than that to which the original data chunk was sent. Thus the multihoming mechanism, supported by machines and network equipment, is a technically feasible and increasingly economic solution [9].

1. 3 .4 Multihoming and Mobility
Note that the multihoming character of SCTP makes it possible to support mobility over IP. Specifically, the SCTP protocol with dynamic address configuration extension can be used to provide smooth handover to terminals (MTs) that pass through different regions of IP networks during an activity session. This is called mobile SCTP (mSCTP) and applies to both IPv4 and IPv6 [4]. Mobility is managed by chunk: Address ConFiguration Change (ASCONF) and Address ConFiguration Ackowledgement (ASCONF-ACK) which offers the possibility to reconfigure IP addresses during the association; Changing the primary route (Routing to a new primary address); Exchanging layer adaptation information during association establishment. These addresses use parameters such as: Add IP Address (ASCONF): add a new IP address to the current association; Delete IP Address (ASCONF): allows to delete an IP address from the current association, etc. 2.2 Our contribution: modeling of the mSCTP protocol using the Markov chain. We highlight the Markov chain to model the mSCTP protocol which constitutes for us the promoter protocol for the mobility management and the aggregation function to show the quality of service (QoS). We consider a cellular network composed of n cells and assume that these are all homogeneous and statistically identical. 2.2.1 Intuitive description model E is the set of states of the system. E is over, and our system is composed of: C= {C 1 , C 2 ,……..C i ……C n } = { 1, 2, …, i, …, n} represents the set of cells or states of the system environment. P ij = P (X t+1 = j | X t = i) is the probability of transition from state i to state j during the time t and t + 1. This is a free system memory for the future state of the system depends only on its present state (this is the Markov property). P t (i) is the probability that the mobile terminal is in the cell C i at the instant t. m is the number of connection (or more specifically named association in the SCTP terminology) between time t and t + 1. mSCTP (t) is a subset of C.
The matrix M t + 1 = [P ij ] t + 1 is a transition matrix (square matrix of order n which models the dynamics of the system transitions from time t to time t + 1). If for any time t, M t = M t + 1 , then the system is called homogeneous (that is a special case of Markov management problems in practice). V= ( P t (C 1 ), P t (C 2 ), P t (C 3 ),....., P t (C n )) = ( P t (1), P t (2), P t (3),....., P t (n)) (5) is the vector position probability at time t of the user in the n cells of the system.

Formal Model Description
is the probability that the mobile terminal is in the cell Cj at time t + 1. Consider a parameter δ, with δ ∈ [0,1], the fixed or variable threshold to select the cells in large probability selected by the mSCTP protocol. mSCTP = {Cj / P t+1 (j) ≥ } (7) 2.2.3 Network Modeling in graph form by the Markov chain The transition matrix of a finite Markov chain can be associated with a graph whose vertices are the states. Pij are the state transition probabilities (i) to (j) (Figure 2).

Aggregation functions for Quality of Service (QoS)
Aggregate functions allow us to provide quantifiable judgment on several intercellular transitions that can guarantee the quality of service. To reach a consensus on these judgments, classical aggregation functions have been proposed: the arithmetic mean and almost arithmetic, geometric mean and almost geometric, median and many others [10,12]. If P ij = P (X t+ 1 = j | X t = i ) is the probability of transition from state i to state j during the time t and t + 1, For n = 4 (number of cells), the graph of the figure 3 we give: P11 = 0; P12 = 0.5; P22 = 0; P23 = 1; P31 = 1; P33 = 0; P43 = 1; P14 = 0.5; P44 = 0. Note also that C1 is the initial state, each state corresponds to an average of random time (0,139s, 0,138s, 0,137s, 0,140s) for data type audio streaming [13] Ti is the transmission rate at the ith associations ⟹ Ti = F (T 1 , T 2 , T 3 , ....., T n ) ∈ [0,1]. T i represents the quality of service (QoS) and F the aggregation function.
T i = ′ ∈ 0,1 ⟹ Q' i+1 ≤ Q i is the quantity of information transmitted from C i to C i + 1 . At this point transmission we assume that the user will receive any new information. If U is the set of information being received in the new cell , then Q i+1 = Q' i+1 + U , with Q i+1 ≥ Q' i+1 which is also the amount of information made available to the user at the input of the second transmission region (2nd Handover ) .

1
Seek the aggregation function that meets our application context [10,12,16] show that the compromise function is located by definition between disjunctive and connective functions and fills most of the properties of the aggregation functions which shows the different transmissions that can be either unidirectional (simplex), Altered (half-duplex), bidirectional (full-duplex), ect. Hence it will be useful for us to better respond to the quality of service in our context.

2 Compromises operators
Two average are entering the contribution of aggregate functions: the arithmetic mean (8) And the quasi-linear mean (9).
Among the two functions (8) and (9), the quasi-linear mean is distinguished by its weight wi (signal strength) which represents the weight of the handover and depends on the altitude between two interfering antennas, Obstacles while verifying the condition: Figure 3 a graph having 4 cells (n = 4) and 5 associations (m = 5). This result is to know in which cell the user is free to move at time t + 1 from the calculation: Finally, the simulation carried out under a Matlab platform is written by the program below. function [Cellules] = cellules Markov(T,Theta,tpn) %Détermination des g(i) g = sum(T'); %Détermination des K(i) K = sum(T); %Détermination des dimensions de T; n:lignes, m:colonnes [n,m] = size(T); %Détermination du nombre d'association nbAss = sum(K); %Détermination du vecteur de probabilité à l'instant 't' ("initiale") Pt = K./nbAss; %Détermination de la matrice de trasition P for i=1:n for j=1:m P(i,j) = T(i,j)/g(i); end end %Calcul des probalités de chaque cellule Cell=Pt*P; for k=2:tpn Cell = Cell*P; end %Détermination des cellules vérifiants le critère "Theta" (les colonnes avec pour valeur "1"

Simulation results in Matlab
We take for example the audio conversion with 0,139s ; 0,138s ; 0,137s ; 0,140s ; 0,145s random time allocated to each state and a constant rate of 40kbit / s for all the cells Ci. We present first the simulation results as a function of f that is an identity function that is, f (x) = x, the weight w and quality of service T( Figure 5). If Di is constant level by : Q '2 =D 1 t' 2 + D 2 t'' 2 with t=t 2 '+t'' 2 and Q 1 known ( 5kbits / s) then T 1 = ′ . We always maintain the audio call data : 0,125s ; 0,130s ; 0,125s ; 0,125s for level 1 and 0,139s ; 0,138s ; 0,137 s ; 0,140s ; 0,145s for Level 2 Q i + 1 = Q' i + U The following combinations are calculated according to figure 3 1 ere association: Q1=5kbits, Q'2= D1*t2 + D2*t'2 with s T1= ′ 2 eme : Q'3=D 1 *t' 3 +D 2 *t'' 3 with T 2 = We present simulation results as a function of f that is an identity function that is, f (x) = x, the weight w and quality of service T (Figure 7).   T= 0,883 ∈ 0,1 Interpretation of 5,7,8,9 figures. We find, in the light of the two first figures, a quality improvement services in the second. We also find that when you change the aggregate function, the transmission rate of the fourth figure provides a better quality of service than the third. That confirms that the second and the fourth figure show some characteristics near reality; for it is rare that the flow rate is constant in a cell (when one moves away from the access point).

. Discussion
Discussed during our studies to propose a protocol to ensure seamless mobility in wireless community networks. Prior Mobile IP (MIP) [1,4] has been proposed to solve the communication problems of rupture during the movement of mobile nodes in IP networks, but [5] we do not understand that it supports fast handoff for applications sensitive delay and packet loss. To resolve this issue, MIP extensions, such as MIP fast transfer (FMIP, fast handover for MIP) and Hierarchical MIP (HMIP, hierarchical MIP), but they do not guarantee a period of minimal notes and a loss of tolerable packages. A second protocol to propose is SIP. In [4], SIP does not provide the transparent management of the handover.]. A third protocol SCTP / mSCTP proposed and thanks to its Multi-homing technique to open multiple IP connections to the same association, and error control mechanism to detect loss, rupture sequences or duplication of packets and dynamic configuration of addresses in an association. The Markov chain have enabled our modeling mSCTP and the aggregation function allowed us to show the quality of service.

Conclusion
It is question for us throughout this article to present the mobility management in own community networks in our communities (Universities, Institutes, School..). We have made studies on various mobility management protocols at the network level, application and transportation while presenting the advantages and disadvantages. These studies led us to propose the SCTP / mSCTP as the promoter protocol of the mobility in wireless community networks, given its advantages presented through the technique of multihoming. the mSCTP mobility protocol is modeled by using the markov chains continuously. Finally researches carried out on aggregation functions show the quality of service on the audio type of application.