A system and method for peer-to-peer content distribution in a weakly connected and heterogeneous personal area network
Original Publication Date: 2008-Oct-15
Included in the Prior Art Database: 2008-Oct-15
This paper describes an automatic peer-to-peer content distribution method between wireless content sharing gadgets in a home environment. There are three important assumptions in such an environment. First, the connectivity between the devices is intermittent. For example, a cell phone might occasionally be in the network neighborhood of a DVR or a music player. Second, there are huge disparities in the content and processing power of these gadgets. The computing power and storage space in a home computer or even a DVR dwarfs that available in a cell phone or a music player. Finally, the connectivity between the devices may be weak. For example, in the popular Bluetooth protocol, the bandwidth is 1 Mbps, much smaller than typical Ethernet speeds. While such network speeds are expected to improve in the future, the amount of content available is expected to rise proportionately to take advantage of the additional bandwidth. This key contribution of this paper is an architecture for the automatic organization and transfer of content objects across heterogeneous and weakly connected devices in order to maximize the availability of frequently accessed content objects in every device. Continuous peer to peer exchange of data is mandated by three factors. First, limited storage capacities on the devices require content to be moved in and out to satisfy different access patterns of the users. Secondly, users can potentially download new content objects on any of the peer devices and they need to be propagated to maintain availability of data. Finally, the rate of content growth is exceeding the increase in storage capacities which implies that the problem is going to get worse in the future. In order to avoid connection latencies and misses due to the absence of requested content, our system fetches content objects in the background without user intervention based on an estimation of future accesses. The content and amount to be transferred between a pair of communicating devices will depend upon the nature of the two devices, resources available for transmission like bandwidth and power, and the criticality of desired data.
THIS COPY WAS MADE FROM AN INTERNAL IBM DOCUMENT AND NOT FROM THE PUBLISHED BOOK
ARC820060095 Jeffrey P Aiello/Almaden/IBM
Prasenjit Sarkar, Shiva Chaitanya, Omer Zaki, Divyesh Jadav
A system and method for peer -to-peer content distribution in a weakly connected and heterogeneous personal area network
Electronic devices are becoming increasingly popular as content sharing gadgets due to the recent advances in storage capacity and network connectivity. Mobile devices are capable of autonomously downloading content via wireless broadband links into embedded storage cards - in the past, such devices had to rely on personal computers with wired broadband links. Furthermore, with the advent of technologies like Wi-fi, Bluetooth and Mobile Ad-hoc Networks (MANETs) , mobile devices can communicate on-the-fly with peer mobile devices as well as larger static devices such as personal computers and digital video recorders. It is envisioned that in the future, more electronic devices such as cameras, camcorders and music players will evolve into content sharing gadgets.
The problem being addressed in this document is the distribution of content across the vast spectrum of content sharing gadgets from personal computers to cell phones. At present, the only way to synchronize content is explicit, manual and tedious in most cases. There are utilities that allow for automatic content synchronization but they are typically personal computer based and specific to a gadget. The personal computer-based synchronization method may be ill-suited for decentralized content sharing networks that wireless gadgets provide. For example, getting an audio file from a DVR to a cell phone could be challenging for the average home user.
The problem of storing large data sets on multiple small storage caches has been well studied in distributed systems [1,2,3,4]. In distributed cooperative caching, systems share data among many caches and minimize the number of hops to locate and access data by placing data close to the clients. The key difference between these systems and the peer devices under consideration is the heterogeneity and weak connectivity. Devices differ in the storage capacities and are often disconnected from each other during normal operations. Mobile disconnected operations in a network file system [5,6] and transparent caching of files based on usage prediction has been researched in [7,8,9,10,11]. The current work differs in the mechanisms by which object popularity distribution information is cooperatively managed by the peers and the optimizations involving data transfers.
2. Summary of Invention:
This paper describes an automatic peer-to-
peer content distribution method between wireless content
sharing gadgets in a home environment. There are three important assumptions in such an environment. First, the connectivity between the devices is intermittent. For example, a cell pho...