Skip to main content
eScholarship
Open Access Publications from the University of California

UCLA

UCLA Electronic Theses and Dissertations bannerUCLA

Near Real-Time Push Middleware

Abstract

ABSTRACT OF THE DISSERTATION

Near Real-Time Push Middleware

by

Siddhartha Mal

Doctor of Philosophy in Mechanical Engineering

University of California, Los Angeles, 2012

Professor Rajit Gadh, Chair

The research conducted in the dissertation is motivated by the needs of near real-time video dissemination to mobile devices and Smart Grid for a near-real-time delivery and filtering system of multimedia content and sensor data. It must also be able to adapt content and data delivery for, and leverage the location determination capabilities of the newest generation of mobile Internet content consumption and generation devices, including smartphones and tablets. A novel near real-time push middleware is proposed, in which end users specify their filtering criteria by creating subscriptions, and pertinent content and data are pushed to them automatically based on their device context parameters, including location. Its function is to receive, store, filter and disseminate content and data from and to mobile Internet devices.

Existing work has addressed the issues of publish/subscribe push data delivery and multimedia content formatting for a variety of target devices. However, the issues of fast subscription matching - specifically location-based, adaptive connection management, and a lightweight client-server push architecture for mobile devices have not been addressed. These problems must be solved in order to realize the aforementioned system. In order to quickly match multimedia content and data attributes to user subscriptions an indexing scheme must be implemented. Solutions exist for standard data (string, integer) indexing, however no adequate method exists for spatial data. A novel in-memory spatial indexing algorithm is developed that is shown to reduce content and subscription search times to a few milliseconds - a reduction of up to three orders of magnitude versus conventional database searches. Client-server architectures exist for mobile data push, however they have high transmission overhead, do not address connection management across varied networks, multimedia content formatting for varied target devices, or device location context. A lightweight client-server push architecture incorporating adaptive connection management is developed which enables client-server push channel connection continuity while traversing a variety of networks and network conditions, and for content to be pushed with very low latency to mobile and traditional clients.

The developed middleware is integrated into two projects: 1) MobiSportsLive, an instant replay video system for mobile clients and 2) a Smart Grid project for electric vehicle smart charging and demand response.

Motorola TuVista was an existing research project that investigated instant replay video creation and distribution to mobile clients. Contributions were made in multimedia content aggregation, reformatting for varied device profiles, digital rights management, and performance characterization. The deficiencies in that project, including a high latency content notification scheme and slow instant replay video creation method, drove the development and integration of the middleware into the MobiSportsLive project. In the MobiSportsLive application the middleware facilitated low latency instant replay video dissemination to mobile clients connected to a variety of networks. It was compared to the TuVista system and content notification latencies were lower by one order of magnitude.

Smart Grid will increase the efficiency and reliability of the electricity grid by adding additional sensing and control capability at the consumer and enterprise levels. Generated data must be collected, processed, and transmitted to appropriate entities for it to drive meaningful near real-time applications. For example, electric vehicle charging sensors generate current, voltage, state-of-charge, and status data. For the regional utility company to be able to monitor and control charging on a broad scale, an intermediate entity must aggregate sensor data from thousands of chargers and disseminate it to the appropriate users in near real-time. To this end, simulations were run in which the middleware aggregated user charging profiles, charger sensor data, alert subscriptions, and demand curtailment signals. The aggregated data was then used to disseminate charging data, vehicle status updates and demand response alerts to simulated clients. The middleware is shown to facilitate electric vehicle charging and low latency alerts and data updates.

The concepts, algorithms, and architectures developed in this research have the potential to significantly improve the state of the art in near real-time multimedia content dissemination and Smart Grid technology.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View