State of Cellular Data Connectivity in India Appropriate Computing - - PowerPoint PPT Presentation

State of Cellular Data Connectivity in India

Appropriate Computing Technologies for Development IIT Delhi Jun 2015

Internet access in India

• Largest growth market, Digital India upswing, leapfrog in economic

development assisted through ICTs

• Barriers to Internet adoption
• Relevant content and services
• User capability
• Affordability
• Infrastructure availability and quality
• Project scope
• QoS obtained for mobile Internet => 2G/3G connectivity
• Mix of rural, semi-urban, and urban locations
• Understand current state of mobile Internet connectivity and any design issues
• Findings prompt us towards the need for better QoS regulation

Measurements methodology

Data collection

• 20 locations spanning rural, semi-urban, and urban areas in MP,

Jharkhand, and Delhi

• Field partners: PRADAN, Vikas Samvad, Air Jaldi, Gram Vaani
• Data collected between 2013-2014

• 1. Availability
• No rural

locations had 3G access

• Wide variation in

availability, most providers have below 80% availability

• Urban locations are

relatively more stable than rural locations

• 2. Throughput (2G)
• Advertised values range

from 144Kbps (EDGE-1) to 256Kbps (EDGE-2, 3)

• Very few locations

approach within 75% of the advertised values

• Fair bit of diversity across

locations, despite high signal strengths ensured in location selection

• 2. Throughput (3G)
• Advertised values range

from 1.4Mbps (UMTS-1) to 2.1Mbps (UMTS-2, 3)

• Less difference from

advertised values than 2G

• Yet, fair bit of diversity

across locations, high deviation from advertised values

• 3. Latency
• Target roundtrip

latency required < 250ms

• Interesting why a variation

across locations exists within the same provider

• Some providers are doing

very well on latency (EDGE-3 and CDMA-1)

Other aspects we have studied

• Page load times for websites
• Hardly any meet the 5sec recommended PLT guidelines
• Website design also matters: AFT/PLT ratios
• Largely a function of latency
• Latency depends on ISP interconnections, backhaul

architecture, CDN presence

• Security
• EDGE-1 and EDGE-2 implement IP spoofing & stateful firewalls,

EDGE-3 and CDMA-1 do not

• Technology issues
• Buffer sizes of 100-150KB are 10x the BDP in 2G, 2x in 3G
• Connection stalls of up to 120sec occur in 40% of flows when

signal strength dips often due to default modem settings

• Mobile users in rural areas run into frequent deadspots with

mean time between failure of 30min

Going forward

• Lots of avenues for technology research
• Regulatory issues
• Framework for QoS regulation: Self reporting, third party auditing of

measurements, nutrition labels, mandatory data disclosure

• Stakeholders: CSOs, web companies, service providers, regulators
• Economic models
• Ownership: Community owned networks like Guifi in Spain
• Extension: Social enterprises like Air Jaldi in India
• Disruptive technology: OpenBTS related projects from UC Berkeley

Thanks!

The ACT4D team

Aaditeshwar Seth, Vinay Ribeiro, Huzur Saran: IIT Delhi Anirban Mahanti, Sebastian Ardon, Siput Triukose: NICTA Australia Zahir Koradia, Arvind Kumar Mahla: PhD students Rahul Nishant, Sameer Agarwal, Amitsingh Chandele, Arvind Subramanium, Sathyam Doraswamy: MTech students Asheesh Sharma, Manveen Kaur: Research Associates Rajesh Kumar, Suresh Mourya, Arun Kumar: Department staff aseth@cse.iitd.ernet.in