Project topics and materials
Get Complete Project Material File(s) Now! »
Context-aware information gathering
The concept of context-aware network selection can be defined as a network selection procedure that selects a target network based not only on the signal quality or explicit advertisements sent by network-side entities, but also on the knowledge of the context information of MN and networks, in order to take an intelligent decision.
In a HWN environment, the selection of the best network is much more complicated than the handover decision between base stations of a traditional homogeneous wireless network. Context information of both terminal-side and network-side are important for making this final decision.
Therefore, a context-aware network selection strategy is necessary, as long as it is with an acceptable complexity. To design a context-aware network selection strategy, a key issue is how to gather all the information, including network context, terminal context, user and application information, etc. All the information should be gathered as soon as possible, because the user cannot wait for a long time for the final decision. Moreover, some information might change frequently, which is an important feature of the information that a context-aware network selection
strategy should consider.
Therefore, a context-aware network selection strategy requires first an architecture of context information management to assure that the information can be available in time. Second, the exchange of information between networks and the terminal should be minimized to save wireless resources. Third, after the gathering of the information, a scheme is required to combine all the information together for making the final decision. In this subsection, we focus on information gathering, and the third step will be discussed in subsection 2.
[BS04] provided a context-aware model which described both static and dynamic detailed information that should be gathered. In [WQ06], an architecture was proposed for context-aware network selection, as shown in figure 2-15. Context information is stored in context information repositories, such as Location information server (LIS), network traffic monitor (NTM) and user profile repository (UPR). Moreover, a handover manager is introduced to filter and process handover-related context information collected from various context repositories. A Service deployment server (SDS) is used to manage and install the service modules needed on both network-side and terminal-side entities. A detailed architecture of context-aware network selection was given by [AT06]. As shown in figure 2-16, this architecture uses the gathered context-aware information for network selection with the following five steps:
• taking user inputs;
• mapping limit values from discrete preferences;
• assigning scores to available networks;
• calculating network ranking based on AHP method;
• session management.
Markovian model
Markov decision process is a common mathematic model for decision making. In the literature of network selection, many studies using Markov decision theory have been proposed. Here, we present several most important proposals on Markov decision theory based network selection.
In [SN08] and [SN07], the vertical handover decision issue was modeled as a Markov decision process, in which the objective is to maximize the total expected reward per connection. The network resources that are utilized by the connection are captured by a link reward function, a signaling cost is used to model the signaling and processing load incurred on the network when vertical handover is performed, and the value iteration algorithm is used to compute a stationary deterministic policy. Besides, this model could take into account the connection duration of various networks, which is an important feature for making the vertical handover decision.
An obvious advantage of this model is the integration of all the above parameters together. In other words, the decision could be made based on just one final decision function which considers all the above parameters. But, according to my understanding, there might be some difficulty in its implementation and testing, because any adjustment of the model will lead to different final decision function. That is not convenient for an engineer to derive again and again different final decision functions. In [SC08], the vertical handover decision issue was further modeled as a constrained Markov decision process (CMDP). Their objective is to maximize the expected total reward of a connection subject to the expected total access cost constraint.
A benefit function is used to assess the quality of the connection, and a penalty function is used to model signaling and call dropping.
According to the authors’ evaluations, this algorithm outperforms the MDP algorithm in [SN08], thanks to the usage of user’s velocity information.
Detailedly, an MT’s velocity is assumed to be correlated in time and can be modeled by a discrete Gauss-Markov random process. The following recursive realization is used to calculate the transition probability of the MT’s velocity.
Game theory
Game theory is related to the actions of decision makers who are conscious that their actions affect each other. The essential elements of a game are players, actions, payoffs, information, etc. These elements are explained as follows:
• Players are the individuals who make the decision. Each player’s goal is to maximize his own utility by a choice of actions.
• An action is a choice of a player as his one-round strategy in the game. For a certain player, he must have an action combination as his strategies during the whole game.
• Payoff means the utility that a player can receive by taking certain action while all the other players’ actions are decided. Therefore, the payoff of one action of one player could change if other players’ actions have any change.
• For each player, there should be a strategy set which contains all the strategies he might choose. In each round, the player chooses one strategy from the set.
• Information, including that of the player himself and that of other players, is important in a game.
• An equilibrium is a strategy profile consisting of a best strategy for each of the players in the game. The equilibrium strategies are those which lead to the maximum payoffs. A Nash equilibrium is a solution of a game, in which no player has any more payoff to gain by changing only his own strategies.
Table of contents :
Acknowledgement
Abstract
Résumé
R.1. Introduction
R.1.1. Réseaux sans fil hétérogènes et la sélection du réseau
R.1.2. Exemples de la sélection du réseaux
R.1.3. Contexte mathématique pour ce sujet
R.1.4. Problèmes à résoudre
R.2. Propositions
R.2.1. Sélection du réseau basée sur la mobilité
R.2.2. Pondération fondée sur des déclencheurs
R.2.3. Une étude simulaire sur la sélection du réseau basée su modèle MADM
R.2.4. Répartition de charge vs. sélection du réseau
R.2.5. Décision HOV basé sur la prévision
R.2.6. Coût de mobilité et sélection du MAP
R.2.7. Stratégie intégrée pour la sélection du réseau
R.2.8. Sélection du réseau pour un réseau mobile
R.2.9. Un simulateur en Matlab
R.3. Contributions
Table of Contents
List of Figures
List of Tables
Glossary
1. INTRODUCTION
1.1. Background
1.2. Problem statement
1.3. Contributions
1.4. Organization of this dissertation
2. STATE OF THE ARTS
2.1. Background
2.1.1. Evolutionary of HWNs
2.1.2. IP mobility and multihoming
2.1.3. Always best connected
2.2. Information gathering
2.2.1. Required information
2.2.2. IEEE 802.21 standard
2.2.3. IEEE P1900.4
2.2.4. Context-aware information gathering
2.2.5. Pilot channel
2.3. Network ranking schemes
2.3.1. Cost/Utility function
2.3.2. MADM
2.3.3. Fuzzy logic
2.3.4. Markovian model
2.3.5. Game theory
2.3.6. NP hard
2.3.7. Power saving
2.3.8. User/operator combined strategy
2.3.9. Integrated solution
3. MADM-BASED NETWORK SELECTION
3.1. Configuration of simulator
3.2. Effects of terminal-side requirements
3.3. Coefficients of various MADM algorithms
3.4. Selection results of various MADM algorithms
3.5. Traffic load assignment for MADM-based network selection
3.6. Important observations and issues
4. MOBILITY-BASED NETWORK SELECTION
4.1. Introduction
4.2. Mobility-based scheme considering two groups of networks
4.2.1. Mobility modeling
4.2.2. Evaluation of HHO and VHO costs
4.2.3. Performance evaluations
4.2.4. Coverage of small-scale networks vs. network selection
4.3. Mobility-based scheme considering N groups of networks
4.3.2. Methods to get the best permutation
4.3.3. Performance evaluations
5. TRUST: A NOVEL WEIGHTING METHOD FOR NETWORK SELECTION
5.1. Modeling the weighting issue for network selection
5.2. Objective weighting methods
5.3. Subjective weighting methods
5.4. Inappropriateness of the eigenvector method
5.5. TRUST
5.6. Performance comparisons
6. MOBILITY SIGNALING COST EVALUATION AND MAP SELECTION
6.1. Mobility signaling cost evaluation
6.1.1. Background on HMIPv6 signaling
6.1.2. Defining a new parameter ‘location rate’
6.1.3. Mobility signaling cost
6.2. MAP selection in HMIPv6 networks
6.2.1. Related work
6.2.2. Location history based MAP selection scheme
6.2.3. Performance evaluation
7. OTHER ISSUES ON NETWORK SELECTION
7.1. Traffic load assignment during network selection
7.2. Vertical handover decision scheme
7.3. An integrated strategy of network selection
7.4. An analysis of network selection for NEMO
8. CONCLUSIONS
APPENDIX
A. Network selection simulator
REFERENCES
