IP P rastogi

Globalization Challenges to Database Community : Globalization Challenges to Database Community Rajeev Rastogi Executive Director Bell Labs Research India


Bell Labs Research India : Bell Labs Research India Launched in October 2004 Focus: Computing & Communication Software Low-Cost Networking Network Monitoring Data Analysis Distributed Computing 3G Wireless Applications Motivation: Access to global talent Access to global markets - India, China: a third of world’s population - New products to penetrate markets Cost savings


Impact, Challenges : Impact, Challenges Impact High-end R&D jobs, better pay Reverse brain drain: many returning to avail opportunities back home Spreading research culture, should boost PhD enrollments Internet enables research to be done from anywhere! New research challenges driven by needs of developing countries Challenges Recruiting Very few quality PhDs graduate each year from Indian universities Defining an independent research agenda Staying connected to US (email, netmeeting do help!) Coping with poor infrastructure: roads, traffic, power,….


Does Globalization really require us to solve new (research) problems? : Does Globalization really require us to solve new (research) problems? My addresses in India Bell Labs Research India, Lucent Technologies India Pvt Ltd., Salarpuria Ascent 3rd Floor, No. 77, Jyoti Nivas College Road, Koramangala Industrial Layout, Ward No. 68, Bangalore - 560 095, Karnataka, India Work 202 Vaswani Exotica, #3 Papanna Street, Off St. Marks Road, Bangalore: 560001, Karnataka, India Home Street name State Zip Country US address format My personal challenge: Getting these to fit into this City


So what else is different about India? : So what else is different about India? 740 million people live in rural villages Low incomes: monthly per-capita income - $17.50 Low literacy: 60% Unreliable power: frequent outages Low teledensity: 1.5 phones per 100 people Low PC penetration Very few Internet users 22 official languages


These lead to new research challenges…. : These lead to new research challenges…. Low-cost computing devices Data access using cell phones Low-cost networks Data access over unreliable wireless meshes User interfaces for illiterate people Multi-lingual information storage, retrieval, and search


Slide7 : Current low-cost PC efforts One Laptop Per Child (MIT Media Labs - $100 PC) Eduwise (Intel) Simputer (Picopeta) Distinctive features Thin-clients (network computing), open-source, low power displays Widespread adoption? Meanwhile… 100m mobile subscribers in India growing at 5m a month! 500m mobile subscribers in China growing at 6m a month! By 2010, 3.7b worldwide (1.5b 2.5G and above) cell phone subscribers! Cell phone costs coming down dramatically (< $50) Could cell phones be the-low cost computing devices of the future? * Source: Pyramid Research Research Challenge 1: Low-Cost Computing Devices


Research Challenge 2: Data Access using Cell Phones : Research Challenge 2: Data Access using Cell Phones Cell phones have many constraints Low bandwidth (few 10s or 100s of Kbps) On 2G phones, communication using SMS messages (max size: 160) Limited memory, battery Possible solutions to conserve bandwidth: Broadcasts [Imielinski etal, Acharya etal] Query: Find dealer offering max price for wheat in neighboring villages Broadcast max price dealer in each village Compute max among neighbors on cell phone Concise answers (e.g. Skylines) Villages Broadcast


Research Challenge 3: Low-Cost Networks : Research Challenge 3: Low-Cost Networks Possible solution: 802.11 mesh networks [TIER, DGP] Low-cost due to competitive mass production Directional antennas for long-distance communication Major challenge: Interference, need clever ways to Assign frequency channels Schedule link transmissions Wireline & cellular technologies too expensive for Internet connectivity Interference


Research Challenge 4: Data Access over Unreliable Wireless Meshes : Research Challenge 4: Data Access over Unreliable Wireless Meshes Data cache Nodes are power-constrained, may be down due to unreliable power Minimize communication to conserve power Cache items based on access patterns Batch queries, route results along steiner tree Route around power outages to maximize throughput DTNs [Fall 03] Data center Data center Data cache Kiosk Queries


Research Challenge 5: User Interfaces for Illiterate People : Research Challenge 5: User Interfaces for Illiterate People Text-based interfaces do not work Need to explore other interfaces Speech Visual (images, video) Benefits Language independent, global Challenges Voice based system needs to learn many accents and dialects Visual interfaces have limited vocabulary (compared to text)


Research Challenge 6: Multi-lingual Storage, Retrieval, and Search : Research Challenge 6: Multi-lingual Storage, Retrieval, and Search Native English speaking population on internet only 35% Databases already support multi-lingual data using unicode Challenge: Support search in many languages over content in many languages Possible solution Translate search keywords to various languages and then search Challenge Need search capability for non-English languages Translations (dictionary, phonetic) need to preserve meaning/intent