From Mobility Management to Connectivity Management

Transcription

1 From Mobility Management to Management Jun-Zhao Sun and Jaakko Sauvola MediaTeam, Machine Vision and Media Processing Unit, Infotech Oulu P.O.Box 4500, FIN University of Oulu, Finland {Junzhao.sun, Abstract This position paper presents the trend of the research focus transition from mobility management to connectivity management, for the emerging pervasive and ubiquitous computing paradigm. Traditional perspective on mobility and QoS resource management is summarized. Scenarios of heterogeneous multiaccess networks are introduced, and new requirements are analysed. Three main tasks of connectivity management are presented. First connectivity is defined as one source of context beyond networking resource. Second features and benefits of flow level vertical handoff are discussed comparing to device level handoff scheme. Third the potential application based on the new conception of connectivity collaboration is presented. A connectivity management framework is outlined in containing the defined functionalities. 1. Introduction With the recent advances in both computing and communication technologies emerging, people begin to think of the new computing paradigms for the future living environment. As a promising research area in information technology, ubiquitous and pervasive computing [1] is attracting more and more attentions of researchers from both academic groups and industrial fields. By pervasive computing technologies, the people s future living circumstance will be penetrated with various embedded or even invisible resources for communication and information processing, where the smart environments under concern can be anywhere including e.g. home, office, meeting room, hotel, restaurant, supermarket, vehicle, to name few. Pervasive computing combines a wide range of research areas as the enabler technologies of this new application scenario. One of the key enabler technologies is communication. It has been widely recognized that the future communication system will base on an all-ip core backbone infrastructure with heterogeneous multi-access networks in order to realize rich accessibility [2]. The scenario that diverse networking resources are converging together to fulfill user and application requirements is mostly due to the great differences and continuous changes of user context in using ubiquitous applications. Next generation networks are appearing with the leading feature of heterogeneous multi-access systems that serves as one of the key enabling technologies of pervasive computing. Pervasive wireless networks are characterized by a variety of wireless overlapping access networks of different technologies and providers. Network connectivity with varying quality of service (QoS) must be offered anytime and anywhere in pervasive spaces. is becoming a reality with new short-range radio systems, wireless local area networks, and packet-based cellular telephone systems. However, effective management of these diverse connection resources within a fully integrated environment is far from maturity. The variety of wireless access networks in pervasive environment introduces a number of challenges for mobility and resource management. Currently mobility management focuses mainly on physical mobile objects such as user and terminal. Traditional network resource management concerns mainly QoS techniques at networking equipments. Moreover, service adaptively based on context awareness is one essential feature of the future intelligent network applications. There is a gap between the current mobility and QoS management technologies and the new requirements of multi-access systems. management focuses primarily on mobility and network QoS mechanisms needed to access ubiquitous services offered through pervasive smart spaces. The paradigm of pervasive networking based on multi-connectivity requires new models and methods both at architecture level and in several layers of the conventional protocol stack. The objective of this position paper is to present the research trend of focus transition from mobility management to connectivity management, for the emerging ubiquitous and pervasive computing

2 Table 1. Mobility and connectivity management Mobility Mgmt Mgmt Non-mobility Portability Location mgmt Single interface mobility Location and handoff mgmt Multiple interfaces, single active Device level vertical handoff context monitoring Multiple interfaces, Crude extension to Flow level simultaneously active device level solution vertical handoff Multiple interfaces, collaboration collaboration paradigm. Table 1 summarizes the key features involved in this trend. The main emphasis of the paper is on the conceptual definition and functionality description of the new capacities of connectivity management. First connectivity is defined as one source of context beyond the original viewpoint of networking resource. Second features and benefits of flow level vertical handoff are discussed comparing to device level handoff scheme. Third the potential application based on the new conception of connectivity collaboration is presented. 2. Vertical Handoff 2.1. Heterogeneous multi-access networks The trend of converging different access systems into an integral and consistent framework results in the future generation network architecture. The top layer is the service and application layer consisting of network data center and various servers such as messaging and web servers. The rest layers are more network related structures, including core and access networks and end user terminals. Access networks are used to accomplish the purpose of really connecting end users to core infrastructure to consume services. The wide ranges of multiple access systems are connected to the core network and interworking with each other. It has been recognized that no one single access technology can cover all the requirements of users and applications. Heterogeneity of access networks comes from many aspects. The access systems may belong to different operators or service providers. Access can be fixed wired, fixed wireless, nomadic wireless, and mobile wireless. Transmission can be both unidirectional and bi-directional. Fig. 1 shows the scenario of a mobile terminal roaming in heterogeneous multi-access networks. From user s point of view, the demand of heterogeneous access systems comes from various user requirements and behaviors. User profiles and Fig. 1. Roaming in heterogeneous networks preferences are different from one to another in terms of device, operator and provider, cost, QoS, etc. To an individual user, application context of communications is continuously changing. From technical viewpoint, the need of heterogeneous access networks is most due to the requirements and situations of different network capabilities and service features. The networking capabilities, in terms of e.g. data rate, coverage, subscriber volume, supporting mobile velocity, anti-interference, suitable transmitting environment, are highly different from diverse access techniques. Moreover, network QoS parameters are dynamically changing from time to time, in terms of reliability and availability, bandwidth, delay, jitter, response time, and packet loss and error rates. Various services and applications request different supports from network and communication systems. Many services have asymmetric traffics in uplink and downlink, e.g. broadcast or GPS positioning. Transmission of multimedia data flows depends on the inhomogeneous distribution of access capacities in the traffic system. The feasibility of heterogeneous access systems is preliminarily depended on the networking capabilities of terminal devices. To be multi-mode or multi band is the basic requirement for device to be connected into multiple networks. Currently the amount of terminals having equipped with more than one network interfaces is increasing, e.g. laptops with LAN, WLAN, Bluetooth, an IrDA, mobile phones with GSM, GPRS, Bluetooth, and IrDA, etc. Autoconfiguration and adaptivity are the abilities needed by future terminals and appliances to automatically discover the environment parameters and optimally configure themselves accordingly Vertical handoff Handoff management is to control the change of a mobile node s attachment point to a network in order to maintain connection with the moving node during active data transmission. Mobile host should be able to roam within a whole mobile communications system or between different systems as long as their networks are interconnected. Networks can be firstly classified

3 according to different providers and/or technologies. Then one symmetric network can be further divided into domains, location areas, access point regions, zones of access points, and logical channels within one access point. Handoff may happen in any of these categories. Different mobility granularities can be defined accordingly. Vertical handoff is the first level of handoff granularity, in which handoff occurs between access points that are using different access network technologies. It is a crucial enabling technology for heterogeneous networks. Typical examples could be e.g. handoff between LAN and WLAN, and between WLAN and GPRS. Horizontal handoff is the handoff between access points of the same type of network technology. This is the traditional definition of handoff for homogeneous network systems. Horizontal handoff occurs with different granularities. Macro-handoff occurs between different visited domains, while microhandoff occurs between different adjacent location areas, access point regions, cell zones, or even logical channels in one cell. It is obvious that the heterogeneity of network technologies brings new concepts into the handoff management, through vertical handoff. First handoff is not necessarily for wireless connectivity only. Fixed access networks like LAN can also take part in the handoff. Next handoff has no direct correspondence with cellular size or the moving speed or range of a mobile device. For example for a location in overlay cells of heterogeneous networks, handoff may occur between the two different cells without necessarily moving out of the coverage area or can be even stationary. This makes mobility a logical concept rather than a physical one, in which mobility just means the change of the logical location of network access points instead of user s geographic position. Heterogeneous networks and vertical handoff is the first cause leading to the research progress from mobility management to connectivity management, which is the scope of the paper. There are mainly two different levels for vertical handoff: device level and flow level. When a device level vertical handoff is performed all the data transfer is switched from one network interface to another. In other words there is exactly one active network interface at a moment. In case of flow level vertical handoff network interface is selected on the basis of traffic flow. Decision on which connection to be used is made for each individual data flow. Simultaneous usage of multiple active network connections is then possible. Obviously flow level vertical handoff is more flexible and beneficial, which forms one of the tasks of connectivity management as described in next section. 3. Management 3.1. Flow level vertical handoff The new scenario of heterogeneous multi-access networks bring many new challenges as well as potentials to network operations with respect to location, handoff and resource management. Current solutions for this situation focus mainly on network layer, by using e.g. Mobile IP to implement upward or downward vertical handoff between heterogeneous access networks [3]. So the connectivity is controlled at device level by using only one active interface at a time. A straightforward question upon the case is that is it possible to use a set of them simultaneously [4], i.e. multiple active interfaces at a moment. Benefits of the simultaneous usage of a combination of network connections include e.g. system can deliver each service via the network that is most efficient and suitable for it, to provide a wide range of QoS to user; one application can have several connections and each traffic flow can select to use the access type that is best suitable to its characteristics each data flow can be dynamically allocated to and redirected between different interfaces to adapt to the dynamically changing network status and balance the workload; access networks may be combined together in use to increase capacity in terms of network bandwidth and reliability, and decrease handoff delay; different access networks might be used for asymmetric traffic, e.g. separate uplink and downlink. With respect to connectivity management there are new requirements on location, handoff and resource management. 1) Location management: In heterogeneous multiaccess networks, one end host may have several network interfaces connecting to various access networks. Mechanisms based on the configuration of one permanent IP address for each terminal (e.g. the home IP address in Mobile IP) is not feasible. A new mechanism is then needed to map the end host name into a set of addresses (i.e. instead of only one address) and further more update the dynamically changing addresses in a group level. Dynamic DNS seems to be an alternative solution to this problem, if updating frequency is not too high and application is not realtime related. Otherwise, new method such as distributed location servers should be employed. 2) Handoff management: The key point of handoff is to keep current network application alive (i.e. no interruption) and to keep current connection alive (i.e. no tear down). In the traditional case of single active interface where one end host may be equipped with only one interface or multiple exclusive interfaces, handoff always happens at device level (i.e. the end

4 Application Application Process Connection and Flow Packet Interface Fig. 2. Handoff at flow level host is seemed as one entity). The terminal can switch between the wireless channels, access points, location areas, or service domains under using. In case of simultaneously multiple active interfaces, higher level handoff operations can be employed, in which the mobile entity can be application, application process, end-to-end connection, data flow, and individual packet. These entities can switch between the diverse network interfaces. For example to application level handoff, all the information exchanges for a particular application use one interface and may switch to another all as a whole. For flow level handoff, a data flow may be dynamically switching between different network interfaces in response to the changing network status. Fig. 2 illustrated the concept of high level handoff with flow level handoff control as the example. Note that according to the new scenario, fixed terminal like desktop PC can also handoff if it has multiple interfaces equipped. Device level handoff is triggered mainly by signal strength. The triggering of the high level handoff should take QoS and context information into account, leading to QoS and context aware resource management as discussed below. 3) Resource management: As one of the most important features of the future pervasive services, application adaptation need to be aware of network context information in terms of QoS parameters. So resource management of heterogeneous multi-access networks must includes the interfaces of obtaining network status by adaptive applications. This is achieved by network monitoring mechanisms. Moreover, QoS and context aware resource management concerns mainly the selection of high level handoff in end host. In other words the operation is to choose one access network to be used for the current application, process, connection, data flow or packet. The selection may happen both at the establishment of new connections and during handoff triggering operation. To decide which interface is currently the best one, some context information should be taken into account, including user profile and preference, application characteristics, and network performance. When a better interface is found, next is to decide whether and when to do the handoff. Handoff includes a series of operations that may bring great overhead in terms of packet loss and delay. A prediction needs to be made on the effect of handoff in order to decide whether it is worth to do so in terms of performance gain. Methods are also needed to avoid the situation of ping-pong effect in handoff. In considering the overhead of connectivity switching, data flow level handoff can be as the compromise choice to balance the granularity of concerned data stream and the cost of handoff operation context There has already been lots of research concerns mobile connectivity, with most of them taking mobile connectivity as a transport resource and so treating it as a communication management issue. Further research takes connectivity as resource context source according to which network-aware applications can be developed to adapt the dynamic changes of underlying network context. However, most of the recent research [5, 6] only take advantages of raw and preliminary network context like bandwidth and develop applications based on the low-level contextual information. Rare research has concerned the integration of context awareness with other related context information and the utilization of connectivity as the source in producing other context information. Positioning system [7, 8] is nearly the overwhelming application in the area, in which connectivity quality serves as the input in deriving the location of the mobile device. is more than just transport resource, nor basic resource context. has more responsibilities than communications and is more potential of providing valuable functionality. In particular in ubiquitous computing connectivity context is an important context info that can be utilized in many ways. Fig. 3 illustrates the connectivity context framework from the viewpoint of context awareness. is collected from underlying infrastructure and service providers, and provided to middleware as well as applications. There are mainly four functions provided by connectivity context manager. 1) It contains the module Collection for the collection of connectivity related raw context information.

5 Storage -Aware Adaptive Applications Interpretation Collection Management 2) It contains the module Interpretation for the interpretation of high-level derived connectivity context data. The high-level connectivity context is deduced through the intelligent interpretation of raw connectivity context possibly with the combination of other contextual information. One example of the deduced high-level connectivity context is best local and remote connectivity for a specific application at the moment, which may serve as the trigger of flow level vertical handoff. Others include time- and location-oriented predicted context. 3) Apart from the direct application of connectivity context like bandwidth in creating adaptive services, connectivity context can also serve as input in the derivation of other valuable context information. As internal utilization interfaces to other context managers are provided so that raw and derived connectivity context data can be used in derivation. Besides positioning there are still many other applications in this scenario including e.g. task distribution, event measurement, location forecasting, user behaviour pattern, etc. 4) As external utilization interfaces to contextaware adaptive applications and other middleware components are provided so that any connectivity data can be obtained and utilized finally for composing adaptive services collaboration Other Interpretation Other Collection Underlying Infrastructure and Service Providers Fig. 3. context framework The selection among and handoff between different active network interfaces for one traffic flow is based on the assumption that the common destination is reachable to all the interface candidates. The idea for this use case of simultaneous multi-connectivity is that traffic flows are independent with each other and are Middleware Components 3G Remote Server UMTS Bluetooth Mobile Device Vending Machine Fig. 4. Vending machine example routed according to their own policies respectively. A further step is that one application may need to open several connections and use them in a collaborative way to achieve an integrated service. In this case the multiple traffic flows are depended with each other and a tight coupling mode is needed comparing to the loose coupling case. As an example, Fig. 4 shows the architecture for an automatic vending machine. The vending machine is equipped with a Bluetooth interface, while the user device with both Bluetooth and UMTS. A Bluetooth connection is established between mobile device and vending machine, and the machine profile is gotten. Some specific software may need to be quickly downloaded from related site through UMTS connection, to smooth the local communication. User browses the information about product availability and price. More detailed information of products can be obtained through UMTS connection to manufacturer s web site to which the address can be locally obtained. A secure UMTS connection is established between mobile device and remote server of customer s financial agent. Electronic credits are downloaded into local device and forwarded to the vending machine to purchase products. The vending machine verifies the credit and then provide product to the customer. Obviously the collaboration mode of multiconnectivity is totally depended on miscellaneous application requirements. Some representative examples can be defined as follows. Splitter, split the entrance and exit for communication, e.g. an outlook application on light-weight client device for management can configure such a splitter that it always receives with GPRS connection while sends using WLAN. Tunneller, directly forward data through tunnel, e.g. a location tracking application can configure a tunneller to forward any input information from GPS connection out to its UMTS interface so that the instant location could be recorded by a remote personal server for later retrieval.

6 Exchanger, switch data packet between connections, in which case the local mobile device simulate a router as the access point for another device. For example a user may configure an exchanger to exchange any packet between IrDA interface and GPRS. Some medium processors can also be inserted. Multicaster, communicate with multiple connections, in case when an application wants to share information with all the parts connecting to it. E.g. a messenger may use a multicaster to send all the output messages to both the GPRS receiver and local WLAN receiver. By collaboratively using multiple connections in both local ad hoc and general infrastructured domains will be the scenario of many future pervasive applications. 4. Conclusions Traditional mobility and resource management technologies are not sufficient when future heterogeneous multi-access networks are under concerning. We analyzed the gap between current solutions and new requirements. We then defined the main tasks of connectivity management in the paper, as the complementary of traditional network management. Three main tasks are discussed in more detail, including flow level vertical handoff, connectivity context management, and connectivity collaboration. A preliminary prototype based on Java on PDA s PocketPC platform has been implemented. Flow level vertical handoff and preliminary connectivity context are already available. More functionality will be implemented gradually during the evolvement of the connectivity management middleware. Moreover, the software platform will further be embedded into a broader platform to experience the investigation of real context-aware applications. 5. Acknowledgement Financial support by the National Technology Agency of Finland is gratefully acknowledged. 6. References [1] M. Satyanarayanan, Pervasive computing: vision and challenges, IEEE Personal Communications, Vol. 8(4), 2001: [2] W. Mohr and W. Konhauser, Access network evolution beyond third generation mobile communications, IEEE Communications Magazine, Vol. 38 (12), 2000: [3] M. Stemm and R.H. Katz, Vertical handoffs in wireless overlay networks, ACM/Baltzer Mobile Networks and Applications (MONET), Vol. 3(4), 1998: [4] X. Zhao, C. Castelluccia, M. Baker, Flexible network support for mobile hosts, ACM/Balzer Mobile Networks and Applications (MONET) 6(2) (2001) [5] B. Badrinath, A. Fox, L. Kleinrock, et. al., A conceptual framework for network and client adaptation, ACM Mobile Networks and Applications, vol. 5, no. 4, 2000: [6] D. Noble, M. Satyanarayanan, Experience with adaptive mobile applications in Odyssey, ACM Mobile Networks andapplications, vol.4, No. 4, 1999: [7] Hallberg, J., Nilsson, M., and Synnes, K. Positioning with Bluetooth. In proc. 10th International Conference on Telecommunications. Vol. 2, [8] Vossiek M., Wiebking L., Gulden P., Wieghardt J., Hoffmann C., and Heide P., Wireless local positioning, IEEE Microwave Magazine, vol.4, No. 4, 2003: