Return to Video Technology Magazine

Economics of Video and the Internet.

John L. Sokol 4/16/03

Hidden costs

The costs of video distribution are often hidden and vary heavily depending on the model used. For most of us today it is difficult to see the business models involved with different video distribution methods. This is because of the technical complexity involved that can make many costs as well as benefits hard to see. I did not want to just write about Internet video without providing a clear understanding the current models that have been successful models so far.

In the Internet boom of late 90’s many companies incorrectly estimated the costs and technical issues of Internet video. Some Internet video vendors actually encouraged these misjudgments. I had some customers who thought that by just placing a video server on the net would start generating revenue, without any marketing web site, or billing system!

In this document I hope to provide a deeper understanding of the business and economic issues of digital video distribution.

Analog to Digital

Today we are at a time of widespread conversion from the older analog systems to all digital systems. The conversion to digital provides many benefits, such as reduced cost, higher image quality, and better audio, at least when it works. Digital compression techniques reduce bandwidth requirements, and allow more channels to fit on cheaper equipment.

The biggest advantage of digital is the use of compression technology, using microprocessors to remove redundancy in the video signal and transmit the minimum needed to generate the end video image. Digital has one big problem though. When faced with a noisy signal it either can fix it 100% or fail 100%, with no in between. Using analog, the picture quality steadily decreases till there is just noise and snow on the screen. In many cases this is more desirable then getting nothing at all. With the conversion to HDTV, many people will get amazing video quality and many who ordinarily get a poor, but viewable TV signal will now get nothing at all!

When I started to work with Internet video it became apparent that we do not have “signal noise” as such, but instead have packet loss. This meant that data we get is 99.99999% correct, if we got the data at all! Loosing at times 20% of the data sent across the Internet was not uncommon at that time. With the early Internet video products loosing over 5% of the data packets would cause no video to ever show up on end users screen. It is a really bad thing when try to run a business and 30% of your customers are not able receive the product they paid for.

Models of Consumer Video Distribution.

For most mediums, whether analog or digital the models are basically the same, with some cost savings associated with digital. Current models include Radio based - Broadcast TV, Cable TV, Satellite TV and Physical Medium exchange such as Video Cassette and DVD rentals and purchases.

Broadcast TV dates back to the 1950’s, the biggest change since then was the introduction of VCR technology. This allowed the broadcast of recorded video instead of having to be constantly on live from a studio with actors and video crew. In this model the operator spends millions of dollars for a FCC licenses, a high-powered radio transmitter on some tall building, and a large production facility to schedule and play recordings and produce live content. I call this front-loaded. It costs $1,000,000 for the first viewer and $0 thereafter, so all the expense is up front. This makes it out of the reach of most people and even companies.

There are derivatives of this, basically renting airtime, such as commercials and infomercials. But the costs are also mixed with the added benefit of a guaranteed audience that was watching something entertaining to hold their interest. Since there is no direct way for them to receive money, selling advertising or as in the case of PBS asking for grants and donations is their only source of revenue. In England the state collects a tax to support state run TV, the BBC. Digital has the potential to change all this, but it doesn’t look likely.

HDTV is the newest thing coming. It allows one TV channel to send 4 regular broadcasts using compression MPEG2 video, or 1 high definition broadcast. In addition the FCC has given every TV station a second channel in UHF space for free with the requirement that they must simulcast their regular program on these channels. This seems like an incredible opportunity, but for some reason they are being slow to move on this.

I can only say one thing to them: “Evolve or die off!”

Cable TV took off in the 1980’s although it dates back earlier. The upfront costs were almost unimaginable running a wire into each and every viewer home. This basically allows a new private set of Radio spectrum. Giving the politics of negotiating with every local government each with there own idiosyncrasies, using their infrastructure of local telephone and power companies, it’s truly a much more impressive accomplishment then most people could even imagine.

For this service they charge some stiff fees of $50 to $200 a month. None of this money goes to local broadcast channels they rebroadcast. Because of the hard wire connection to each viewer home, there is a higher sense of security and increased quality to the video transmission, although neither is necessarily true.

In summary the up front cost was astronomical and the cost per viewer is around $100 for a tuner, or now days of a digital receiver that decompresses the video. There is also a constant maintenance cost of the infrastructure similar to that of the phone system. It’s no wonder that they are looking at proving alternate services over the cable network such as Internet and Telephone server. At this point though the cable companies have completely botched subscriber music.

The cable business model is interesting since they do not produce content at all and are just providing transit for other peoples content. They are more of a marketing company reselling basic and pay channel services, keeping a large share of that revenue. Then there are the content providers such as CNN and HBO that sell their feeds over regular satellite to cable providers. Some also include advertising as well. They do not have transmitters and don’t even need an FCC TV broadcast licenses but instead have a Satellite uplink, $200,000, and have to rent a satellite transponder at around $100,000 a month.

Satellite TV dates back to the 1950’s also, is interesting because in its analog form it was only used internally to the industry. For consumers who could get analog receivers in the late 1970’s, viewing satellite TV was a theft of service in its very nature, snooping in on what ever transmissions were passing by. Most satellites could only carry a few analog channels 36 Mhz wide. Today even these are compressed and encoded digitally and encrypted, pretty much killing analog satellite TV. It’s only with Digital compression that subscriber Satellite TV services became available in the early 90’s. Hughes pioneered the technology of having an all-digital satellite and allowed DirectTV to take off. By using digital with compression and error correction, receiver dishes became much smaller and cheaper with all of the cost going into the digital receiver. The up front costs again were millions of dollars; the cost per viewer is a $100 digital receiver that is very similar to the cable TV Digital receivers. Over all the rest of the model is almost identical to cable TV. With one exception, catching people for theft of service is almost impossible since there are no wires to each location so they have become much more depended on digital encryption and security smart cards to prevent theft.

One other model that is very different is physical tape and DVD disks. In this case the video has been converted to a physical medium and can be rented and sold like books or magazines are. A company like Blockbuster has the substantial overhead of a large inventory and rapidly depreciating product that is almost like a food product. Most movies lose value over time; there is a kind of shelf rot with these products. It appears that they have cut a special deal with the Studios to deal with this where they pay a percentage of the rental fee. This allows them to pay for the inventory of the plastic DVD or tape but not the virtual inventory of it content. Tape being worth 50 cents while the move may retail for $80. Viewers need to have a DVD or VCR for viewing costing from $100 to $400.

Other applications/business models of video.

Besides the mainstream consumer models there are a number of other video applications that are commercially successful. Some are internal to the video industry such as Satellite video links, editing equipment, recorders, mixers and other stuff you’d expect to see at the NAB (National Association of Broadcasters) trade show.

Other applications involve everything from machine vision for parts inspection in factories, to casino security systems that record hundreds of cameras constantly looking for thieves and cheaters.

DVR’s and CCTV surveillance systems alone could fill an entire book. Later I would like to show how these systems could benefit tremendously from the advances in desktop computers and the Internet.

In analog systems the VCR has become a universal component for recording video in many applications. Today the DVR (Digital Video Recorder) is starting to replace this role, but there are few standards. Many used non-standard controls, compression and transmission mechanisms to reduce cost and take advantage for newer technologies that the standards haven’t caught up with yet. This includes the Internet where many people would like to treat it as just a direct video link. No product that I know of can do this effectively, although it is technically feasible. But the Internet can be so much more than just another transport medium like a satellite or microwave link.

There are also some applications that seem great but have just not taken off.

Here is a list of some applications that I keep hearing about over and over again.

Home and Business video conferencing – video phones.

Cell phone video conferencing. 3G, 4G , PDA and set-top-box conferencing

Multipoint video conferencing, with more then 2 sites.

Medical Video Conferencing. Telemedicine.

Education – remote educations live and recorded. Also called video education. Education in the classroom. Remote classrooms.

Remote engineering support, connection factory with engineers remotely to diagnose a problem.

Pay Per video videos over the Internet. (Hollywood movies, independent films, news, sports, live psychics, religion, Adult / sex) the last one seems to be the only successful one.

Set Top Video boxes over the LAN, Internet (Intertainer for example)

Hotel room video rental.

Live event broadcasting. This is something I pioneered with Sun Microsystems and Hazardous Media.

Live Event uplinks to live TV Broadcast; we say a lot of this from CNN with the war in Iraq.

360 Deg. Viewing of video. (Inward or outward) inward would be like an avatar or stadium, where outward would be like QuickTime VR. Where the viewer is on the inside looking out

3D Stereo video compression, transmission and recording.

Telepresence, Remotely controlling something over a video link. An AUV or Arial Autonomous vehicle would be an example that allows the operator to control an airplane remotely like it were a video game.

Video answering machine. Video E-mail. Video intercoms. Virtual portholes.

AR Augmented reality. Where image recognition. Beacons, and GPS connect to databases to provide real-time data overlaid with real world activates.

Personal DVR, record everything you see during the day.

Remote tuner viewing. For example watching San Francisco cable TV from Korea.

Live Interactive Video presentation, like PowerPoint with Video. Also with conferencing

Video surveillance, remote viewing, remote archiving.

License place character recognition for auto-toll bridge, gate access, highway vehicle tracking, Automatic speeding tickets based on travel time between 2 points.

Tracking a person’s movement throughout a building.

Steaming Internet video infomercials, or commercial financed video content.

I cannot cover all these concepts in just one document.

Most are cool, but many will never be commercially viable or successful.

All of these models can be put into a few conceptual categories.

Point-to-Point – 1 sender to 1 receiver. Live and Recorded Content

Video Conferencing is an example, also microwave links.

Broadcast - 1 sender to many receivers. Live and Recorded Content

Each receiver sees the same image at the same time. TV and Cable.

Local Recorded –- 1 recoding per 1 viewer VCR tapes and DVD’s.

– Hard Drives (TIVO) Downloaded movies on the internet.

VOD - Video on Demand - Transmission or 1 recoding to 1 viewer, but many viewers at different points in the recording. Similar to Point-to-Point and Recorded but play live while transmitting. Hotels pay per view. Korea Video rooms

Each of these operates best on different transmission mediums.

Radio – UHF, VHF, Microwave, Satellite, Cable TV Coaxial cable, laser.

These are one Broadcast video per a given channel. Where a channel is a 6Mhz block of RF Spectrum. Even though digital is more efficient it is still limited to the same rules.

Radio is inherently a broadcast medium. With the exception of directional microwave and laser links, creating a point-to-point transmission over radio sacrifices part of a limited spectrum to that conversation. To resolve this cell phones break an area up into cells each with it’s own radio tower transmitting with much lower power levels that reduce the area covered. To make up for this, the cell system must then use many radio towers, each one only able to cover few miles radius.

Directional microwave and laser are point-to-point and not broadcast, but are line-of-site only. These tend to have a very involved to set up. Sprint Broadband was one of few companies to offer microwave Internet access, but stopped after 2 years of service, probably because it wasn’t profitable.

Wireless radio is by its very nature noisy and unreliable. Signals can interfere with each other and signals can be reflected back interfering with itself or get blocked by hills or valleys, bad weather and solar activity. This causes dropout and data errors and data loss, which is a problem for digital video transmission. Cable TV, which uses wires, mostly avoids these kinds of problems, but isn’t totally immune.

Phone Lines – Modems, ISDN, DSL, T1, T3, OC3 Fiber Optics. (Leased Lines)

These are called transit in some circles, since they just move data and do not necessarily connect to the Internet.

All of these are point-to-point, and are very inefficient as a broadcast medium.

Most of these are extremely reliable for data transmission, but in a packet switched environment such as the Internet, will have problems when they become congested.

Ethernet / LAN – these are almost always connected to a point-to-point system (WAN and/or Internet) but can be broadcast within a building or campus using Multicast or broadcast packets.

Each medium is efficient for it’s model, but trying to broadcast on a point-to-point medium or provide point-to-point on a broadcast medium is very difficult and inefficient.

Physical Recorded Medium – VHS Tapes, DVD, in Bulk the cost can be as low as 25 to 50 cents each. In single quantity is around 1 dollar per tape and DVD-R disk; there is also the VCD format that allows regular recordable CD’s to be used in DVD players.

These are about 25 cents each in single and can get incredibly cheap with bulk-manufactured disks. Not count the effort of having to copy or burn the video on to each disk by hand. From a consumer point of view there is something tangible with they purchase a DVD that they don’t get when downloading a file over the Internet. The general attitude of people today is that files on a disk don’t possess any value, they can be freely shared or traded or deleted. Any new system of selling video will need to overcome that mentality in the consumers. On the other hand, people often will spend $20 and buy a DVD when they can just record it off Cable TV that they already pay $100 a month on.

Internet Technology

By this I mean TCP/IP Networks, IPV6 but most of everything here will apply to any other packet switched technology, such as Frame Relay, ATM, MPLS networks and the like.

The Internet can carry all data types and run on top of all communications media! This is a big statement but it’s true and has to do with how it evolved. IP works over radio, fiber optics, satellite, serial ports, even PC printer ports, USB, Ethernet, Modems, can be tunneled though any other networking technology. It can even be recorded to tape or disk and played back for things like multicast video transmission where network conversations are one way.

What this means is the Internet is the Rosetta stone of digital communications, allowing computers to transparently communicate over a dizzying array of seeming incompatible networks.

One mistake make with the Internet was in thinking it could be commoditized on the basis of bandwidth, and considering all bandwidth of the same capacity as being equal. The reality was the quality and reliability of the different Internet connections varies greatly and this was is an easy thing to quantify. For many applications were speed doesn’t matter as much such as mail and web pages, this is usually fine, but for video and audio this is critical. For people pioneering video over the Internet were always the first to find network problems and had to be the most discriminating and critical of the networks quality, where quality being measure by uptime, packet loss and latency.

.

Terms:

Bandwidth - Data Rate - The amount of information required per second to play video. It can also mean the total capacity of the local Internet connection although this is seldom obtainable. The Internet connection bandwidth needs to be at least slightly larger then the required bandwidth of the video transmission.

Latency - The delay from the time a message is sent to when it is received. (From Korea to US is 150 ms each way or 1/6^th of a second. (300 ms round trip). Also referred to as PING time. When a network has problems these can rise to 2 seconds or more, effectively killing a videoconference.

Errors/Loss - The data is corrupted or lost in transmission, With most modems and Internet we never get to see errors, the equipment will delete or correct them so all we get is lost packets. These then need retransmission and increase the latency while the client wait for the missing data to arrive. Statically the retransmission can also get lost, so TCP/IP will exponentially drop data rate with packet loss. A packet loss of 5% can kill the possibility most TCP/IP based video transmissions. Loss is most often cause by congestion on part of the transmission route over the Internet.

Uptime – is the amount of time the Internet connection is working correctly. It’s actually a measure of how much the connection fails completely. These days uptimes are good, usually 99% or better. Even today though, a misconfiguration, virus or worm, DOS (Denial of service) attacks can take the network down or cause sufficient loss to prevent video streaming and conferencing. DOS attacks are the worst since an attack on a neighbor’s computer can also disable your connection as well.

These 4 terms define the scope of the transmission problems.

Looking at these terms for different recording and transmission mediums.

Medium	Loss	Latency	Bandwidth
Local recording	0	0	High > 1Mbps
Broadcast Radio	Errors*	0	High > 1Mbps
LAN	0	0.1ms	10Mbps to 1000Mbps
Leased Line	0	30ms	64K to > 1.5 Mbps
Modem	0 to 1%	50ms	2.4K to 56K
Internet	0.1 to 20%	30 to 500ms	Depends on connection
Packet Radio	0 to 20%	> 100ms	1.2K to 128K depends on service

* Broadcast radio can vary greatly depending on antenna and location.

Of these transmission media, the Internet can be an order of magnitude lower in cost to use, in particularly for international transmissions where distance is not considered when billing.

The cost of a dedicated international T1 is greater then $15,000US a month.

However the cost of 2 T1's, one in each country with Internet is usually less then $3000US a month.

Note: The Internet access in some countries is insanely expensive. Sri Lanka in 1998 was imposing a $24,000US a month tax for each 64Kbps of bandwidth!

The Internets Hidden Cost.

When people use the Internet they pay a monthly internet bill typically $50 USD for DSL or a cable modem or $10 for a dial-up modem over the phone lines. Content providers, such as someone who wants to distribute video over the Internet, generally pay for bandwidth usage or are capped at a fixed maximum bandwidth.

When serving data over the Internet, the client at home only pays for ½ the cost and the server with content pays for the other ½ of that cost. This sounds innocent enough but when scaling up this will kill you. Each guy at home say, uses 10 cent per hour, but since he pays a flat rate and so is insensitive to the cost of his bandwidth usage. So he can leave the video running all day without incurring any additional cost. For the server though, 10,000 viewers would cost $1000 per hour $24,000 per day or $720,000 a month. ($720 per megabit per month and assuming each client gets 100Kbps of video) The actual cost depends on the bandwidth of each video stream and the pricing deal that can be negotiated with the ISP, but in general this is a typical price but could run much higher. With 10,000 viewers this is small potatoes compared to broadcast TV with millions of viewer. In one year 8.6 Million dollars you could almost buy a TV or cable channel. There are ways around this, but this is what killed many idealistic bright and well-funded startups in the dot.com boom. Certainly one way is to make sure you collect at least more then 10 cents per hour of video per user.

I will go into this more specifically later.

Internet as a Broadcast Medium.

The Internet, which is comprised mostly of point-to-point lease line connections and LAN’s can be used to broadcast. Broadcasting requires a change from the traditional way of doing things.

A look at the cost of Internet verses broadcast TV.

Cost breakdown for sending video.

In this graph I am not concerned with absolute costs but the cost trends with the number of viewer.

Broadcast TV

As I mentioned earlier, there is a large initial investment and large cost to reach the first viewer and costs are essentially free after that.

Internet streaming

In a typical internet video/audio streaming solution a single server or multiple are placed at one location and connected to a large leased line, or placed in a facility called a CoLo (Co-Location site) where several large leased lines provide access to the internet. The initial setup costs are relatively small, a few thousand dollars, compared to millions for a TV Stations. The bandwidth required is the data rate of an average video stream * the maximum number of viewers * 1.05 (this is + 5%) for additional overhead.

This type of video over the Internet can only compete when there are only a low number of viewers. The bandwidth costs skyrocket very quickly with more viewer.

Distributed Serving

This is sometimes referred to as a rebroadcast network, edge servers, or caching servers. Examples of this would be Broadcast.com, real.com, Akamai. With distributed serving many low cost servers are paced on cheaper, lower quality bandwidth, but since there are many servers users can be automatically connected to the best possible server for them. This combined with the power that a large amount of internet traffic brings allows them to start negotiating “peering agreements” these are essential free or low cost internet bandwidth because it would start costing the backbones and large ISP’s money to have them deliver the content delivered off of someone elses network. So ISP’s see moving cheap bandwidth as a benefit by saving them money and proving better service to their customers. Basically there is a lot of politics and some very fancy marketing and negotiations on the part of the content rebroadcast companies. This can effectively reduce the ½ cost the content providers were responsible to carry.

Peer to peer

This was given a bad rap from the fight between the music industry and napster. But napster’s popularity and large audience is a tribute to its technical and economic model.

This is the only model where the total cost of bandwidth usage is covered but the clients who receive the content. Users not only receive their music and video files but also re-share them to other users so they will on average send as much as they receive. Not very popular with the ISP’s since the cost to ISP’s start goes up, but this is just passed back down to the end users, so they end up paying 100% of the bandwidth they are using.

In the case of napster this was just files, the content was provided ad hoc by end users who could very easily just ignore copyright and post what ever they like with anonymity.

Napster’s only cost was the initial software development and the bandwidth costs of the central database that allowed end users to find each other. The central server was like a dating service, connecting two automated software programs together, allowing a PC with data to connect with a PC that wanted that data directly to each other.

But Peer-to-Peer could also enforce payments, copyright and protected and encrypted information. It can transmit not only recorded files, but also live streaming content. Pay per view movies and live events could be delivered from set top boxes and desktop PC’s on DSL and Cable Modems using this technology for a minimal cost.

Peer to peer could one day kill the video rental store and music companies or it could transform them to a more profitable and efficient state then ever thought possible. It just depends on how they choose to embrace change or avoid it.

The COST of INTERNET BANDWIDTH

Bandwidth is measured in several ways.

In average/peak bandwidth usage or total amount of data sent. Bandwidth is directly corresponding the maximum video viewers on a system so it’s a much more usable measure for people serving video.

To convert from one form to the other::

1 Mbps (megabits per sec) / 8 =

125K Bytes per second. * 60 =

7.5 Megs per minute * 60 =

450 Megs per Hour * 24 =

10,800 Megs (Megabytes) per day =r 10.8 Gigs per day.

Approx. 300 Gigs per month

So an average of 1 Mbps = Approx. 300 Gigs per month served.

Now the amount of data send is always changing and it is send in packets. Packets are small groups of Bits from 40Bytes (320bits) to 1500Bytes (12000 Bits) always sent at the maximum data rate of your connection. The speed is determined by the timing between packets sent. With TCP/IP the speed is decreased when packets are lost.

When talking about packet loss, averages are always referred to, taking into account the delay (empty space) between packets. Typically averages are taken over a 5-minute, hourly, or daily period. Over the course of a day the peak hours can be 20 times that of the low usually at night. Also during peak a server may see a drop in data rates being served This is a clear sign of a poor quality connection from the ISP or backbone connection that is starting to drop packets coming in over congested network links.

Most ISP’s bill for the upper %5 of peak bandwidth usage or in gigabytes per month served.

At this point it’s about $1 to $4 per gig served. Or $600 to $1000 a month for an average of 1Mbps served. Also many make you choose a peak usage with a hardware device to hard enforce that limit creating severe packet loss when that level is exceeded limiting the hardware, or imposing steep rate increases for exceeding that level.

In the USA this cost can vary greatly depending upon the connection quality and city. But as a base line price for a good connection averaging 1 Mbps cost around $1000 US per month this is $1 per Kilobit per sencond per month.

This could include short bursts peaking of 100 Mega Bits per second if you ISP and server are capable of it without incurring an additional fee.

Also the price is about constant for T1 1.54Mbps at $1500 US per month.

ADSL at $200 a month for 384K or a cable modem at 1+Mbps for $50US is extraordinary! But both have problems during the peak in the evenings, especially the cable modems that can become totally unusable at certain times of day. Also they are asymmetric meaning they can receive at higher data rates, but are usually limited to 128Kbps or lower for uploading. In addition DHCP is used to allow the ISP to move your IP address around on the fly. The effectively prevents people from serving data from their low cost home connections.

One of the cheapest and also the lowest quality bandwidth is a PPP modem account at $10 to $20 per month for 56Kbps modem. Also not practical for serving data, although in a peer to peer environment programs like napster have still been able to utilize these connections for sharing large files of several megabytes each.

The COST of VIDEO.

Often I have had people think compression will change the cost of serving video. In practice though you want to send at much data as is available to a low bandwidth connection. So the cost and requirement of sending video is more dependent on the type of connections the customers are connected in with and not the compression algorithms.

At around 100Kbps video looks somewhat natural and can complete with a poor VHS recording, meaning you might actually be able to sit through a 2 hour movie encoded in MPEG4 or DIVX formats.

Cost for a 28.8 Modem user then costs us $30 Per month of viewing VIDEO!

Cost for a 56K modem = $60 a month, so for a 56K viewer. 1 Day cost $2 US and 1 Hour Cost approx. 8.33 cents just under 9 cents per hour!

I know the numbers here don’t agree with the numbers I reported earlier. The truth is the prices really do vary by a factor of 4 from ISP to ISP even within the same city.

NOTE: The Internet is unreliable though! For the Worst Case multiply the bandwidth by 2 for strong error correction or many retransmissions.

You still pay for the data even if you must retransmit it a second time.

For a really screaming video connection at 500Kbps cost

$500 a month = $16 Per Day = $0.70 Per Hour. = 1.2 cents per Minute!!!

A compressed telephone audio channel is about 1 cent per Hour!

CD quality Stereo Audio (128Kbps) for MP3 (MPEG3 Layer 3 audio) is 18 Cents per Hour. And for a stereo voice conference like that 36 Cents per hour. Again on a fixed rate connection like DSL or cable modem, these costs are never seen.

NOTE! For conferencing multiply the one-way numbers by 2 , one for each direction.

Compared to a SATELITE VIDEO UPLINK or VYVX at over $1000 per Hour

For Professional Quality video = 30Megs Bits per second (light compression)

= $30,000 per month = $1000 Per day = $42 Per Hour. (Worse cast with Error Correction $100 per hour)

Cost of Mobile Wireless Internet.

Regular Internet was around $1 to $4 per gig, that’s 0.0004 Cents per Kbyte

Verizon wireless charges 5 Cent per Kbyte. That about 10000 times the cost of regular Internet connectivity. No where on there site could I get an indication of speed, or what data rate they could send, but they had a Logitech video conferencing package for these so it must be around 28Kbps. Too bad Metricom went out of business since they had a 128Kbps wireless modem Internet access for $60 a month.