The goal of the research effort in NIMBLE is the design, evaluation, and deployment of information services using the concept of Infostations. The architecture consists of a ubiquitous low bandwidth wireless network which is augmented with convenient and frequent access to isolated high bandwidth connections which we call Infostations . This network supports a ``many-time, many-where'' communication paradigm that is suited for a wide variety of information services. Different kinds of Infostations are being deisgned:
Sit-through Infostations
Walk-through Infostations
Drive-through Infostations
An additional aspect of this work is our consideration of ``asymmetric'' transmission to achieve very small, very low cost terminals having very low energy requirements.
A dataspace is a massively distributed database that strongly couples the physical space with the information space. This coupling is achieved by requiring all producers of information store to their owndata locally. Proactive requests for information by users of the dataspace are answered by querying the physical location in which the desired information is produced. Dataspaces shift elementary database operations closer to the transactions that normally take place in the underlying network. In effect, the network itself becomes a vast database system supporting the querying and monitoring of billions of distributed information sources.
Infodispensers (Digital Sprinklers)
Infodispensers are data aggregation and distribution devices that blanket the physical landscape and act as information sprinklers, sprinkling important local information on users as they pass nearby. They do this by caching frequently queried data from the local dataspace and dispersing it to passing users based on the users' previously specified information preferences. Infodispensers make it possible for individuals to passively receive information of interest to them without initiating an explicit dataspace query.
The Spatial Web builds on the infrastructure of Infodispensers. It provides a mechanism for organizing and representing information aggregated by infodispensers. As with traditional pages on the World Wide Web, spatial web pages contain hypertext links to other pages. However, spatial web pages are different from standard web pages in two important respects: (1) all spatial web pages posses a limited physical "datascope", and (2) the links within the spatial web page mirror the relevant structure in the physical space. The spatial web facilitates the retrieval of spatial information by allowing its users to "crawl" the physical space in which the information is stored, in much the same way as they crawl the information space using traditional web technology.
DBMate is a project investigating the effects of mobility on database systems, especially in the context of disconnected client server systems. This includes study of how the physical organization of the database server can be tuned for mobile clients, how consistency issues can be resolved when disconnected clients perform local potentially conflicting updates and how quality of service issues apply to mobile systems. Some of these concepts developed as a part of DBMate are now being applied to the WEB& project.
The goal of this project is to support disconnected operation for the Web. This project is designing an architecture for non-interactive web. We have a preliminary prototype that allows background web queries. We are aiming for a comprehensive directory service, uniform service descriptor, transaction support, and service composition.
IP multicast is an efficient point-to-multipoint distribution mechanism. However, there are a number of scenarios in which a reverse, multipoint-to-point aggregation mechanism is highly desirable. We introduce a programmable mechanism, called gathercast, to support the aggregation of packets without altering any of the routing or forwarding mechanisms of the Internet. Gathercast is based on active services framework and can be deployed incrementally. It works well within the current IP multicast model. We have implemented it in our own network. One of the aggregation mechanisms that we study in this paper is the combination of small packets using gathercast. Small packets constitute a large fraction of packets in today's Internet. Every packet requires a routing table lookup and incurs the same performance cost irrespective of the size of the packet. Gathercast allows creation of an Internet HOV (car pool) lane, in which multiple small packets to the same destination are combined into a larger packet. Web servers, in particular, are a leading cause of small packets. All the TCP ACKs generated by the clients have the same destination (the web server). These ACKs are just 40 bytes in size. Another example of small packet traffic is ICP (Internet Cache Protocol) queries generated among web proxies. ICP queries are about 60 bytes in size. Gathercast combines such small packets into a larger packet (and regenerates the original packets at the destination), thereby reducing the number of packets seen by routers in the network. Simulation studies show that gathercast-enhanced ICP scales with the number of proxies, and reduces the the number of generated packets by a factor of 16. Simulation studies also show 11% to 68% reduction in packet loss and 2% to 30% improvement in performance of busy web servers when used in conjunction with gathercast.
A mobile user invariably encounters a heterogeneous network consisting of segments with diverse properties (bandwidth, asymmetry, reliability, etc.). We need the ability to modify the packet flow over various segments of a route for adapting to such properties. For such adaptations, we propose the concept of transformer tunnels. Transformer tunnels transform the packet flow to provide application-transparent, route-specific adaptations. Route-specific adaptations, unlike adaptations that are forced on all hosts using a link, allow different hosts using the link to simultaneously request different adaptations. The adaptations depend on the transformation functions attached to the tunnel. We have implemented the transformer tunnels and have used them over our wireless network to provide adaptations for wireless links. In this paper, we show how mobile hosts can use transformer tunnels to change packet flow over the last-hop link. We also provide an API for adding new transformation functions to the system. Using this API, we have implemented some transformation functions such as encrypting data to provide security, sending packets in bursts to allow energy efficient operations, combining small packets and compressing data over slow links to improve link throughput, and so on.
The current internetworking protocols are limited by their logical view of the network because they cannot relate the physical geographical world to the internetwork. Yet, mobile wireless users will have need to send, receive, and access information that is pertinent to the user's location. With the advent of the Global Positioning System (GPS) and the availability of chip-sized GPS receivers [GPS], all future mobile wireless nodes can be equipped with the knowledge of its location. A user's location will become information that is as common as the date is today; getting input from GPS, when outdoors, and other location providing devices, when indoors. Availability of location information will have a broad impact on the application level as well as on network level software.
Some of the possible new services and functionalities include geographic messaging, geographic advertising and services, and geographic querying. For example, sending an emergency message to everyone whom is currently in a specific area, such as a building, train station, or a highway.
To support such applications, location should become a first class citizen, starting from the IP level and proceeding all the way to the application layer. Routing protocols for geographic messages need to be developed. Furthermore, the geographic destination for a message should not be confined to a single point, but should be able to be specified as any arbitrary polygon.
MRVP is a reservation protocol for mobile networks. In order to reduce the impact of mobility on the QoS guarantees, a mobile host has to make advance resource reservations at several locations it may possibly visit during the lifetime of the connection. The currently proposed reservation protocol in the Internet, RSVP, is not adequate to make such reservations for mobile hosts. In this project, we have designed and implemented a reservation protocol, MRSVP, for providing real-time service support to mobile hosts in an integrated services network. MRSVP is an extension of the resource reservation protocol, RSVP, and it can make spatial resource reservation for a mobile host.
The availability of variety of communication devices offers a choice among networks with vastly different characteristics. No single protocol or application can be expected to perform well over all these networks. A mobile host is likely to encounter these different networks and needs to adapt accordingly. The problem of adapting to a changing network environment is further complicated, because changes in network conditions are usually transparent to higher layers of the protocol stack. In order to allow automatic adaptation of applications and protocols, awareness of link conditions and network environment is necessary. In this paper, we present a uniform mechanism based on ICMP messages for providing information about the environment to the protocol stack. We also show how protocols can adapt to changes in the environment, and in particular, demonstrate dynamic fine tuning of some of the well known protocols such as UDP and TCP. Performance measurements demonstrate that our framework imposes little overheads and improves protocol performance under changing network conditions.
The present web has a shortcoming that although it is successful at describing conceptual space, it does a poor job of describing physical space. Let us say, for example that we are interested in answers to queries such as "Find the closest Chinese Restaurant with the minimum waiting time", "Inform me whenever this freeway becomes congested within two miles or less", "How many allied troops are within a one mile radius?". The current web architecture does not support such location dependent queries, depending on dynamically generated data from various kinds of sensor networks. Obtaining such rapid, accurate and useful information in response to location-dependent questions as these would provide a whole new level of convenience and empowerment to individuals who ask them. This project aims at defining the architecture, protocols and services to facilitate monitoring of the physical world.
To provide Quality of Service(QoS) to flows, routers need to maintain per-flow states. This would require that all the routers in the Internet be changed to provide QoS. Any approach, which reduces the number of routers to be changed, will probably be more feasible than an approach requiring all routers to be changed. This project investigates edge router based approaches in this regard. The goal of the project is to design an edge router based framework to emulate the services provided by an Integrated Service network.
Last
modified: February 10, 2000
Maintained by Samir Goel <gsamir@cs.Rutgers.edu>