Mobile Web Service Discovery in Peer to Peer Networks

The advanced features of today’s smart phones and hand held devices, like the increased memory and processing capabilities, allowed them to act even as information providers. Thus a smart phone hosting web services is not a fancy anymore. But the relevant discovery of these services provided by the smart phones has became quite complex, because of the volume of services possible with each Mobile Host providing some services. Centralized registries have severe drawbacks in such a scenario and alternate means of service discovery are to be addressed. P2P domain with it resource sharing capabilities comes quite handy and here in this paper we provide an alternate approach to UDDI registry for discovering mobile web services. The services are published into the P2P network as JXTA modules and the discovery issues of these module advertisements are addressed. The approach also provides alternate means of identifying the Mobile Host.

💡 Research Summary

The paper addresses the emerging problem of discovering web services that are hosted directly on mobile devices such as smartphones. While modern hand‑held devices now possess sufficient memory, CPU power, and high‑speed cellular connectivity (3G/4G) to act as service providers, the traditional approach of publishing service descriptions (WSDL) in a centralized UDDI registry becomes impractical. The authors identify three core shortcomings of a centralized registry in the mobile context: (1) scalability – the sheer number of services that could be offered by thousands of mobile hosts would overwhelm a single point of registration; (2) dynamism – mobile nodes frequently change IP addresses, join or leave the network, and thus require constant re‑registration, leading to stale entries; and (3) reliability – a central server constitutes a single point of failure and a bottleneck for discovery traffic.

To overcome these issues, the authors propose a fully distributed discovery mechanism built on the JXTA peer‑to‑peer (P2P) framework. JXTA provides language‑agnostic, XML‑based “Advertisements” that describe peers, groups, and services. By mapping a mobile web service’s WSDL information into a JXTA Module Advertisement, each Mobile Host can publish its service directly into the P2P overlay without needing a public IP address. The paper details how the lightweight JXME (JXTA for J2ME) client enables resource‑constrained smartphones to act as Edge peers in the overlay.

The network topology is organized into three logical peer types: Edge peers (the mobile devices themselves), Relay peers (which traverse NATs and firewalls), and Rendezvous/Super peers (which index advertisements and forward discovery queries). The authors suggest deploying a Super peer at a mobile operator’s base transceiver station (BTS). This Super peer combines the functions of a Relay and a Rendezvous node, allowing mobile devices to register using only a Peer ID and their phone number, thereby eliminating the need for a globally routable IP. The Super peer also propagates advertisements across multiple BTSs, creating a virtual P2P network that spans the operator’s entire coverage area.

Discovery proceeds as follows: a client device issues a query (by service ID or keyword) to its local Rendezvous peer. The Rendezvous peer consults its local index and, if necessary, forwards the request to neighboring Rendezvous peers via Relay nodes. Matching Module Advertisements are returned, containing the WSDL URL and binding information. The client then directly invokes the service using standard SOAP over HTTP/HTTPS. When a Mobile Host disconnects or changes its network attachment, its advertisement is automatically withdrawn, preventing stale entries. Conversely, when the host obtains a new IP address, it republishes an updated advertisement, ensuring the overlay remains consistent.

The authors implemented a prototype Mobile Host on a Sony Ericsson P800 (Personal Java) and a JXME edge client on three additional smartphones. Performance measurements showed an average end‑to‑end service response time of roughly 1.2 seconds, with transmission delay dominating the total latency. The distributed discovery added less than 10 % overhead compared with a hypothetical centralized lookup, and the system scaled linearly as more peers and services were added. The paper also surveys related work (JXTA‑SOAP, JXTA‑WS, etc.) and argues that most prior efforts target desktop environments, whereas this work explicitly tackles the constraints of mobile devices and integrates the solution with existing cellular infrastructure.

Despite the promising results, the paper acknowledges several limitations. The reliance on XML‑based advertisements and SOAP messages incurs considerable bandwidth overhead, which may be problematic on low‑bandwidth 3G links. The current discovery mechanism supports exact‑match ID lookups but lacks sophisticated keyword or semantic search capabilities needed for large service catalogs. Security considerations are only briefly mentioned; the system would benefit from a robust peer authentication and access‑control framework (e.g., PKI‑based certificates). Finally, the deployment assumes that mobile operators are willing to host Super peers at BTS sites, which raises operational and cost concerns.

Future research directions suggested include: (a) integrating lightweight protocols such as CoAP or JSON‑RPC to reduce payload size; (b) extending the advertisement schema to support richer metadata and semantic annotations for more expressive queries; (c) designing a distributed trust model that can authenticate Mobile Hosts and clients without central authorities; and (d) evaluating the approach on newer 4G/5G networks with higher throughput and lower latency to assess scalability under realistic carrier conditions.

In summary, the paper presents a viable alternative to centralized UDDI registries for mobile web service discovery by leveraging JXTA’s decentralized advertisement mechanism, adapting it to the constraints of smartphones, and embedding the P2P overlay within existing mobile operator infrastructure. The experimental prototype demonstrates that service discovery can be performed with modest latency and acceptable overhead, laying groundwork for future, more secure and scalable mobile service ecosystems.