Selected Papers by Ravi Jain

Future telecommunications networks will consist of integrated packet-switched (IP and/or ATM), circuit-switched (PSTN) and wireless networks. Service providers will offer a wide portfolio of innovative applications over these integrated networks. Doing so rapidly and efficiently requires open network APIs, with a key API being that for call control as well as coordination and transactions. The JAIN community is defining an API for Java Call Control (JCC). The JCC API defines the interface for applications to initiate and manipulate calls. Note that in this context a call refers to a multimedia, multiparty, multi-protocol communications session. The JCC Edit Group of JAIN has defined the first version of the JCC API specification, which has been released in January 2001. This chapter describes the basic motivation and design of JCC 1.0.

The Mobile IP (MIP) protocol for IP version 4 as being standardized by the Internet community provides continuous Internet connectivity to mobile hosts, without requiring any changes to existing routers and higher-layer applications. We propose an alternative protocol, Mobile IP with Location Registers (MIP-LR) which is closer to the "service node" database approach used in the Public Switched Telephone Network (PSTN): before launching a packet to the mobile host, the sender first queries a database to obtain the recipient’s current location. MIP-LR is designed for operation in enterprise environments or within logical administrative domains, as it requires a sending host to be aware which hosts are potentially mobile and implement the MIP-LR protocol. The benefits of MIP-LR are that potentially long routes, called "triangle routes", from the sender to the mobile host are avoided, encapsulation of packets sent to a mobile host is not required, the load on the home network as well as the home and foreign agents is reduced, and there is substantially improved interoperability with protocols such as RSVP for providing QoS guarantees. We carry out a simplified average-case analysis of the costs and benefits of MIP-LR and show it can result in significant reductions in mean network costs compared to MIP.

Abstract. The Mobile IP (MIP) protocol for IP version 4 provides continuous Internet connectivity to mobile hosts. However, currently it has some drawbacks in the areas of survivability, performance, interoperability with protocols for providing QoS. We have proposed an alternative protocol, Mobile IP with Location Registers (MIP-LR), which overcomes some of these drawbacks and is closer to the "service node" database approach used in the Public Switched Telephone Network (PSTN): before launching a packet to the mobile host, the sender first queries a database, called the Home Location Register (HLR), to obtain the recipient’s current location. MIP-LR is designed for operation in enterprise environments or within logical administrative domains.

In this paper we focus on showing how MIP-LR enhances the survivability of MIP by eliminating some of the functions which MIP introduces for mobility support, allowing the HLR to be placed outside the mobile’s home network in case the latter is particularly vulnerable, and replicating and distributing HLRs. We present two schemes for managing the multiple HLRs and enabling mobile and correspondent hosts to dynamically discover the addresses of the HLRs serving a given mobile host. The first scheme introduces a set of Translation Server (TS) databases while the second uses a form of quorum consensus based on the Triangle Lattice (TL); for the latter we present an enhanced protocol called the Optimistic TL (OTL). For both schemes we present algorithms for mobile host registration and packet delivery, protocols for recovery from HLR failures, and complexity analysis of the overhead involved.

The WWW has made information and services more available than ever before. Many of the first Web applications have been emulations of real world activities, in particular, e-commerce. But so far, the use of information and services on the Web has been a solitary one. We propose a component-based architecture for collaboration that provides shared navigation of the WWW along with an EJB-based server implementation. As a particular application built on this architecture, we present MultECommerce, through which multiple users can participate in virtual shopping trips among multiple shopping sites. MultECommerce features a multi-site shopping cart and enables one-stop checkout from all visited shopping sites. We examine security and performance issues of our architecture.

In next-generation enterprises it will become increasingly important to retrieve information efficiently and rapidly from widely dispersed sites in a virtual enterprise, and the number of users who wish to do using wireless and portable devices will increase significantly. This paper considers the use of mobile agent technology rather than traditional client-server computing for information retrieval by mobile and wireless users in a virtual enterprise. We argue that to be successful mobile agent platforms must coexist with, and be presented to the applications programmer side-by-side with, traditional client-server middleware like CORBA and DCOM, and we sketch a middleware architecture for doing so. We then develop an analytical model that examines the claimed performance benefits of mobile agents over client-server computing for a mobile information retrieval scenario. Our evaluation of the model shows that mobile agents are not always better than client-server calls in terms of average response times; they are only beneficial if the space overhead of the mobile agent code is not too large or if the wireless link connecting the mobile user to the fixed servers of the virtual enterprise is error-prone.

We study the efficacy of using link-layer retransmissions to improve TCP performance over lossy wireless links. The scenario we consider is where TCP packets traverse a wired network to a base station, and thence over a single wireless hop to a stationary receiver. Unlike many previous studies, which rely on simulation, we develop an analytical model for calculating TCP throughput; unlike all previous analytical studies, we do not ignore the possibility of acknowledgement packets (ACKs) being lost on the reverse link from the wireless receiver to the base station. The analytical model captures the performance of four common TCP algorithms: OldTahoe, Tahoe, NewReno and Sack.

We find that, for the scenarios studied, a moderate number of link-level retransmissions can significantly improve the throughput of TCP and its resilience to packet losses on the forward and reverse links; increasing the number of retransmissions further has relatively little benefit. We also show that ACK losses cannot be ignored unless the number of retransmissions permitted is high or errors on the reverse link are infrequent. Of the TCP algorithms studied, we find that Sack has the best performance, closely followed by NewReno; the performance of OldTahoe is very poor. We end with suggestions for further work.

Future telecommunications networks will consist of integrated packet-switched (IP and/or ATM), circuit-switched (PSTN), and wireless networks. Service providers will offer a wide portfolio of innovative applications over these integrated networks. Doing so rapidly and efficiently requires open network APIs, with a key API being that for call control, as well as for coordination and transactions. The JAIN community is defining an API for Java Call Control and Java Coordination and Transactions. The JCC API defines the interface for applications to initiate and manipulate calls, while JCAT defines the facilities for applications to be invoked and return results before or during calls. Note that in this context a call refers to a multimedia, multiparty, multiprotocol communications session. The JCC/JCAT Edit Group of JAIN is in the process of defining the first version of the JCC/JCAT API specification, which is expected to be released in early 2000. This introductory article describes the background of and motivation for the design of the JCC/JCAT API. We briefly describe the AIN and JTAPI call models on which the JCC/JCAT API is based. We then describe the scope of JCC/JCAT and its relationship to other JAIN Edit Groups defining facilities for enabling service creation. Finally, we describe the requirements and example service drivers for JCC/JCAT, as well as the initial proposed design and structure for JCC and JCAT.

We present the architecture, design and experimental research prototype implementation of ChaiTime, an open system architecture for the rapid development of advanced next-generation telephony services that overcomes some of the limitations of the current closed PSTN architecture and service model. ChaiTime allows communication sessions to be set up over the PSTN, the Internet, or a combination of both. Services can be provided by multiple cooperating distributed service providers, some of whom may use third-party software components which can be "plugged in" or even dynamically downloaded from the network as needed. This allows advanced services to be deployed and delivered to users rapidly, a crucial requirement in the increasingly competitive telecommunications services marketplace.

ChaiTime is built upon an object-oriented call model called Java Call Control (JCC) which we have defined as a small set of extensions to the standard Java Telephony API (JTAPI) call model that allows support for distributed providers as well as advanced services. JCC hides details of underlying call-state management, protocols and hardware from applications. In our prototype, we have designed a small set of extensions to SIP, called Extended SIP, for supporting advanced services. The ChaiTime prototype software is currently operational in our laboratory. We briefly describe its current implementation as well as future work to address issues such as fault tolerance.