Other Free Encyclopedias » Online Encyclopedia » Encyclopedia - Featured Articles » Contributed Topics from P-T

Placement of Continuous Media in AD-HOC Networks of Devices - Mobile Devices

data clip storage clips

Definition: Using the ad-hoc network, we deliver data from the local storage of one or more neighboring devices to reduce the demand for the network infrastructure to remote servers. Thismode of delivery requires the devices to collaborate with one another by sharing a fraction of theiravailable storage.

One may enhance availability of a clip by bringing it closer to the device that displays it. Toelaborate, in , it was noted that the overall bandwidth required to implement an interactivevideo-on-demand solution based on a naive design that employs one centralized server wouldbe as high as 1.54 Petabits per second for the entire United States. Using the ad-hoc network,we deliver data from the local storage of one or more neighboring devices to reduce thedemand for the network infrastructure to remote servers. This mode of delivery requires thedevices to collaborate with one another by sharing a fraction of their available storage. Inreturn, the storage manager provides physical data independence which means the physicalorganization of data can be modified without causing application programs to be rewritten. Itempower authenticated users to stream a clip to any device as long as it has either wired orwireless network connectivity to devices containing the referenced clip. This means thesystem (instead of the user) resolves the identity of the device that delivers the data requestedby the user. Moreover, the system may offer each user a larger amount of storage capacitythan that offered by one device. The exact capacity is dictated by the total storage of devices connected in the ad-hoc network and the capacity of remote servers. This storage might be shared by service providers that provide households with on-demand entertainment content.

Bringing clips closer to a user means that an individual who employs others’ devices to share his or her experiences may have personal content (typically a fraction of it) pre-staged onmany devices in anticipation of future access. This may raise a host of privacy, copyright, andlegal issues. While there are techniques for some of these challenges, a significant amount offuture research is necessary to ensure privacy of users’ personal libraries. We believe theadvantages of physical data independence will usher-in a new host of techniques to addressthese challenges.

Placement of data consists of a collection of techniques to: 1) collect statistics about theenvironment and how the application references data, 2) place data across devices, 3) reorganize placement of data in response to changes in access profiles and the environment.

Each topic is vast and most research to date has focused on topic number 2. Formally, given arepository of C clips with a pre-specified frequency of access (f;) for each clip i, a dataplacement strategy addresses the following questions. First, what is the granularity ofplacement for data (a block or a clip)? Second, how many replicas of a granule should beconstructed in the system? Third, how should these replicas be placed across devices?Answers to these questions are a trade-off between (a) the average startup latency and (b) thenumber of simultaneous H20 devices that can display clips. These two metrics constitute thedimensions of a recent experimental study used to evaluate three data placement strategies foran ad-hoc network of stationary H20 devices: Simple, Halo-Clip, and Halo-Block.Granularity of data placement is a clip with both Simple and Halo-Clip. It is a block withHalo-Block. Simple employs the profile of a device to pack its storage with its mostfrequently accessed clips. If the demographics of each device is identical then Simple assignsthe same collection of clips to each device. Halo-Clip strives to maintain a replica of the clipsthat constitute the repository across the devices participating in the ad-hoc network. This ispossible when the storage capacity of the participating devices exceeds the repository size. Similar to Halo-Clip, Halo-Block also strives to maintain a copy of every clip in the ad-hocnetwork. When sufficient storage is available, it replicates the first few blocks of each clipaggressively in order to enhance startup latency.

Mobile Devices

In, we investigated how many replicas of different clips should be constructed in a mobileenvironment that provides on-demand access to audio and video clips. This environmentconsists of vehicles equipped with a Car to Car Peer to-Peer (C2P2) device that might serveas a component of the vehicle’s entertainment system. Mobility is the key difference betweena C2P2 and a H20 environment. With C2P2 devices, a vehicular entertainment system offersits user a list of available movie titles during the car’s journey. A particular title is availableonly if sufficient replicas of that title are expected to be encountered in the vicinity of the carto enable successful viewing. However, a title may have a certain time delay after which it isavailable. We define a related QoS metric for content availability, termed availability latency, defined as the earliest time after which the client vehicle encounters a replica of its referenced title. To minimize this metric, the system may replicate popular clips more aggressively than the less popular clips.

A family of replication techniques to compute the number of replicas for a title as a power law function of its popularity, i.e., frequency of access, is presented in The exponentvalue (n) identifies a specific technique. Three distinct exponent values are studied in :random (n=0), square root (n=0.5), and linear (n=l). Availability latency is impacted by alarge number of system parameters such as density of C2P2 devices in a geographical area,title display time, size of clip repository, trip duration, the mobility model, storage per C2P2device and the popularity of the titles. We refer the interested reader to for details.

In , we explore the use of mobile C2P2s that carry a referenced data item from a mobileC2P2 containing that data item to a client C2P2 that requested it. Such devices are termedzebroids. A device acts as a zebroid when it is in close vicinity of a server C2P2 and travelsalong a path that rendezvous with the client C2P2. A key finding of is that zebroidsenhance the availability latency of a client with a random mobility model. An investigation ofthis finding with other mobility models is a future research direction.

Planck, Max (Karl Ernst Ludwig) [next] [back] Piven, Jeremy - Actor, Career, Sidelights

User Comments

Your email address will be altered so spam harvesting bots can't read it easily.
Hide my email completely instead?

Cancel or