Understanding Last Mile Connectivity

Continuing to my last post about why broadband is so poor in India.In last post I discussed little bit about “last mile”.

There were few confusions, let me first clear them out before I end up topic about last mile.Someone left comment on last post and said The pvt. telcos really think we Indians are idiots are idiots and don’t need high speeds.come on yaar give 8mbps @ 999 with no FUP and see the rush for BB.

Very good confusion!

Let me clear this one. Firstly how many people you think know the difference between 2Mbps and 8Mbps, and how many really need that much speed? Me, you and other folks here do understand what is 8Mbps and what one can do with that, we will be happy to get one, but are we in majority? Infact a study shows top 5% user on broadband network constitute for 57% of resource usage. Why not ISP’s should cut those 5% and make life easy for 95% users?

Next thing, most of us already have broadband and we pay close to 1000Rs a month for that. Point is how many you think are still not having broadband because there’s no [email protected] ? If ISP comes with that, they won’t really get new people signing up for service, so what’s the profit?

Airtel is already giving 50Mbps, BSNL is offering 24Mbps for 7000Rs a month. Main reason for such high pricing is the fact that there’s no business model which supports lowering prices. Don’t expect something super rocking from small players as they do sit on Domestic & International bandwidth of big players. They can’t kill of business of big guys sitting on their backbone! (Do you think billion $ companies are that much stupid?)

Next (good) question was = Why low demand is keeping rates high?

Well, reason is broadband is quite different from things like say electricity. Bandwidth is virtually unlimited (as oppsed to case of electricity), but to create that sort of “unlimited” capacities we need heavy infrastructure consisting of fiber links, routers, switches = making backbones, datacenters followed by servers serving that much data on the network. Now if less users use it, it takes cost per user quite high. As more and more users come up, cost/user comes down.

Other question is confusion about wireless broadband. Is wireless broadband cool thing? What about 3G & WiFi?

I would again say here – 3G is good if you come from path like 2G, 2.5G and 3G. It’s very good as for fast connection while on go, but it was never designed to be an alternate of fixed broadband. Reason remains “limited spectrum”. At the end, you have limited spectrum available which carries data. It’s true that HSPA+ can give a real world speeds of 5-6Mbps, but overall capacity is limited, and cannot offer “everyone” with those speeds. That’s why 3G is a good option for low demand areas like Tier 3 cities, Rural areas etc. I myself stay at Rural area during weekdays and I have seen EDGE performs better here as compared to EDGE in New Delhi.

Other part here is = at the end 3G is “just a last mile” technology and it needs significant backhual capacity to provide users with broadband. Most of mobile towers are still running over point-to-point microwave and that itself has limited capacity followed by varying latency. It works well for low bandwidth applications like voice calls but it will be a challenge to provide high speed broadband over 3G over such backhual capacity.

Next point = wifi. Many people have confusions about WiFi & it’s power.

Well, the big reason for success of wifi is fact that it operates in unlicensed spectrum 2.4Ghz. Now unlicensed makes it available for everyone (not like exclusive spectrum which telecos carry). This makes wifi prone to heavy interference. To cover a city like say New Delhi with wifi, it’s a big technical challenge. Firstly, you can’t build big towers like you have for cell phones, because big tower = covering more area = covering more users = too much interference apart from limited 20Mhz spectrum in 2.4Ghz band. And a Ghz band has lot lower penetration then Mhz band (on which telecos operate).

Way out is = small wifi hotspots. To atleast cover whole area with wifi coverage effectively, we need something around 1 hotspot after every 100m. Nearby hotspots will operate on different (non-overlapping) channels of wifi, and will enable 20Mhz of bandwidth per hotspot. This way is again = good for wireless/handheld devices where you tend to consume “decent” amount of data in sending emails/chats/ a couple of videos etc. It still can’t give a 8Mbps to each home user, enabling them to download 2-3GB of data in a day.

In real = most of projects which started to provide city-wide wifi in Western countries = failed. Technically it seems like a non-feasible idea, apart from fact when you build hotspots after every 100m, you need a wired backhual to feed those access points, apart from constant power feed (which is again not easily available) creating economical challenges. WiFi hotspots will become more popular in crowded areas like Airports, Malls, theatres etc of big cities as here aim is mainly to offload peak traffic from cell phone networks. Aim of such wifi network is not to give a super high speed broadband on cheap rates (to home users), but to provide a high speed connection for limited usage on go.

Bottom line = there seems NO wireless technology available till now which can give you high speed broadband experience and is scalable. You will get point about scalability once you will see choked 3G networks after few months!

Other confusion = capacity of fiber optics.
Many people feel fiber optics offer almost unlimited capacity and we can do anything. It’s half correct. To understand capacity on fiber optics, understand it as “free open space” outside. Now to run a car/bus smoothly on that “big space” you need to build road. Fiber is much like that. In peak rates, you can have 10Gbps/wavelength of bandwidth over single-pair (2cables, one for sending and one for receiving). Now you can have 160 wavelenghs (peak) over such pairs uding DWDM. This takes capacity to as high as 1.6Tbps over single pair. Most of fiber pairs deployed on backbone are like 100-200pair fibers and this takes theoretical peak bandwidth to as high as 100Tbps. But here one big thing is = fiber is just a media = just a part of network. There are lot more components required in such networks, which are still bottleneck. BSNL’s core network is still at just 10Gbps, and they are upgrading it to 40Gbps. Just an upgrade from 10Gbps to 40Gbps needs over few thousand crore’s of investment over new equipment (other then fibers). So in theory = yes fiber has almost unlimited capacity, but in real it’s limited due to economical factors.

In case of Verizon FiOS (one of biggest fiber to home deployments in US) they give 50-100Mbps of bandwidth. They can surely take it high to 4-5Gbps, but to offer end users that much bandwidth, they need to upgrade backhual capacity and that needs more investment.

Here’s a summary of (popularly) available last mile options:

Name of technology Type Good Bad
DSL Wired (twisted pair phone lines) Has enough bandwidth to serve existing needs for India. It seems OK atleast till end of 2012. Only BSNL+MTNL hold most of copper and their “unions” are not permitting Govt. to unbundle last mile for other operators. For new players, DSL doesn’t seems like long lasting due to limited bandwidth.Distance is big limitation of DSL After 1Km distance from exchange, it’s gets too noisy to provide high speed broadband.
Cable broadband Wired (co-axial cable) Has lot more bandwidth then DSL. Using technologies like DOCSIS 3.0, ISP’s can easily provide broadband speeds of 50Mbps over coaxial. In India there’s hardly any big cable tv operators. Super high fragmentation limits their size, apart from fact most of last mile co-axial is badly managed and doesn’t supports two way Internet. We need totally new infrastructure, and in age of fiber, big guys tend to ignore such heavy investment.
3G Wireless over 2100 band (with recently allocated) 5Mhz spectrum One of very cheap deployments for broadband. Just needs investment in upgrading equipment on existing towers and increase in backhual capacity. In low demand and low ARPU’s it can surely help. Allocated spectum is way too less for most of broadband needs. 5Mhz is way to less and can’t provide everyone with good broadband speeds in big cities.
Mobile WiMax Wireless over 802.16e over recently allocated 20Mhz spectrum Better then 3G technologies. Is more scalable, and good enough to support decent need of mobile broadband. BIG problem here is limited spectrum. Cities like New Delhi need over 200Mhz spectrum, while allocated is hardly 10% of that. Apart from that, globably WiMax seems failing agaist it’s competitior LTE. Thus overall ecosystem is also a problem for WiMax.
WiFi Wireless over 2.4Ghz on 802.11n. Global standard, and can be used on almost all laptops and most of new handheld devices. Spectrum is open and thus free for ISP’s to use for deployment. No heavy spectrum fees. Open spectrum makes it terribly prone to interference, apart from limited bandwidth of 22Mhz. To cover big area, wifi needs lots of Access Points & to serve them itself is a challenge.
GPON Optical Fiber Extremely scalable, and supports super high bandwidth. Technically one of best available technologies for broadband deployment. Needs completely new infrastructure, and it’s costs a lot. Low ARPU makes it very hard to deploy such technology. Apart from that cost of termination electronics is yet to come down. Slicing fibers, connections, etc all needs trained technicians as opposed to current low-trained “linemans” and that also makes it expensive.

Well my post seems again going too big to cover next part. Will cover rest in next post 🙂

Comments are always welcome!

About the author:

Anurag is a Student and a part time network admin at dito.

He is all around network related technologies like DNS, Internet routing, servers, connectivity etc. He is also one of Power Posters at Official Google Apps forum and is involved in deployment of Google Apps.

Anurag is doing bachelors in Information & Technology from a State college in Haryana.

Reported By

Leave a Reply

93 Comments on "Understanding Last Mile Connectivity"


Sort by:   newest | oldest
Abhishek Gupta
January 19, 2016 6:00 pm 6:00 PM

Hi Anurag
1. Need help on delivery of MPLS and Internet link on same RF last mile both in case of SDH network and CEN (Career Grade Ethernet Network) network.

2. What shall be the LAN port requirement in case of SDH and CEN network for delivery of both the link? Will it require single LAN port or 2 LAN port? Or it will be different in both the cases? As per my understanding, in case of SDH min. 2 port will be required whereas in case of CEN single LAN port will suffice.

3. Please suggest how it will be achieved.

Vinay Chandran
March 8, 2014 11:20 pm 11:20 PM

Hi Anurag,

Need some advise on last mile options for enterprise MPLS. Fibre, FSO or RF?

Need a comparison if possible

Thanks & regards,

Vinay Chandran

March 1, 2013 3:13 pm 3:13 PM

What you will say about Internet Lease Line,MPLS…..services……

Anurag Bhatia
March 20, 2013 1:51 am 1:51 AM

What about them?

Basically they have last mile over copper, fiber and wireless based on whatever is available to nearest customer location. Also usually enterprise clients enough on their bills that last mile can be managed specially in metro areas where getting fiber to one building gets you many new clients.


Subscribe to our email newsletter