Wireless Networking for Home and Small to Medium Businesses
By Yoram Solomon,
Vice President & General Manager, Advanced Communications Business Unit
PCTEL, Inc.
Abstract:
In the first part of this article, I will analyze the home and SMB usage of the Internet. I will introduce reasons to have broadband connectivity that are not typically considered to be reasons to have a broadband connection. These reasons will include the time it takes to download web sites and the overall quality of Internet usage. In addition, I will analyze the bandwidth consumption over time.
In the second part of the article, I will show how a broadband Internet access combined with a home network will give all users of the home network the best performance, almost regardless of the number of users using the shared Internet connection.
In the third part of the article, I will compare the different home / SMB networking technologies including: wireless, phone line, power line and Ethernet. I will mention the different standards in each technology. I will conclude the article with a discussion of the value proposition that a home and SMB network must meet in order to enter the mainstream.
Introduction
The traditional way of connecting to the Internet today is still by a telephone line with an analog modem, which provides connection rates of up to 56 Kbps. While broadband services are offered such as DSL, cable, and fixed wireless, overall new subscription rate is still low. On the other hand, there seems to be an interrelationship between subscribing to a broadband Internet service and implementing a network, especially in homes and small to medium businesses. The number of computers per household is growing beyond a single computer. With the most important value proposition of a personal computer today being Internet connectivity, there will be more than one user wanting to use an Internet connection at the same time. Once a broadband connection is available, all Internet users want to use it, thus making it attractive to install a home or small office network to share that link. On the other hand, if a network is already installed to share files, printers, schedules or other resources, getting a broadband connection starts to make more sense even though the price is higher.
In this article, I will discuss the Internet usage models in homes and in small to medium size businesses, describe the issues and benefits that arise from combining a broadband Internet connection with a network. I will then move on to review the different types of available networking, and conclude by defining the value proposition required for wide deployment of wireless home and SMB networks.
Part A - Internet Home and SMB Usage Models
On February 16, 2001, the U.S. General Accounting Office (GAO) published a report titled “Characteristics and Choices of Internet Users”. According to this report, there are over 100 million Americans online. 89% of users rated e-mail as ‘very important’ or ‘extremely important’. 78% of users said the same about web surfing. Less than 35% attached the same importance to audio and video material. Other Internet uses such as e-commerce, chat rooms, and personal web pages seemed to be even less important. When considering Internet users, we may identify three general types. I will refer to them in the following paragraphs as the ‘Communicator’, the ‘Researcher’ and the ‘Media Hog’.
The Communicator
The main purpose of the Communicator is to stay in touch with other people, and e-mail is the main Internet application of the Communicator. In additional to e-mail, the communicator sometimes uses chat rooms, newsgroups and instant messaging. These applications typically consume very low bandwidth, and delays in the transmission or reception of information cause little inconvenience to this type of user.
The Researcher
The Researcher surfs the web for information, and uses search engines to find the required information on the Internet. They will read information either on the screen, or print out to read hard copy. One interesting point to note is that the human mind cannot read and digest information provided at 56 Kbps. In other words, even with a dialup Internet connection, we can download information faster than we can read and digest it.
The Media Hog
The Media Hog is typically a teenager who downloads music, video clips and computer games from the Internet. These types of data have a totally different ratio of data-rate to experience time. An MP3 file of a 5-minute song will be approximately 3 megabytes. It will, therefore, take almost 9 minutes to download with a perfect 56 Kbps connection – more time than it takes to listen to it! Downloading the same song with a 384 Kbps DSL connection will take 1.3 minutes, and using a 1.5 Mbps connection will take only 33 seconds. The media hog frequently uses the most bandwidth available and truly appreciates it.
Why have a broadband connection?
The Media Hog undoubtedly needs a broadband connection. He or she downloads music files from Napster, watches short Internet movies on ifilm.com, and takes advantage of the bandwidth to its fullest. The media hog can really appreciate the difference between downloading a Metallica song through a 56 Kbps connection versus downloading it through a 384 Kbps (or more) DSL connection.
For most people, 56 Kbps dialup speed (and lower) is sufficient for the “hot” applications listed in the GAO report. E-mail, instant messaging and web browsing do not really mandate the high-speed connection. In fact, the report shows that while 52% of Internet users in the U.S. were able to subscribe to broadband Internet services (either DSL, cable, or wireless), only 12% actually did. The reasons were that speeds afforded by regular dialup were good enough for their purposes, and their current usage patterns could not justify the higher price of broadband Internet. When dialup service is offered at $19.95 per month and broadband services are offered between $39.95 and $49.95, users have difficulty justifying the difference in price.
“Time is Money”
Website Download and the Overall Quality of Internet Usage
Although at first glance, connection speed does not seem to be important to the Researcher, the typical web surfer or even the Communicator, it does affect the quality of their Internet access. When designing a web page, a designer is typically directed to limit the size of all objects on the page to whatever can be downloaded in 10 to 20 seconds using a 56 Kbps modem. Assuming most web pages are designed according to this guideline (keeping in mind that many pages use flash animation or other bandwidth-consuming objects), moving from a 56 Kbps to a 384 Kbps or higher connection speed means that the same page can be downloaded seven or more times faster. For the Communicator, sending and receiving e-mails with large attachments such as pictures or documents will be faster using a broadband connection, which in turn leads to a more pleasant user experience.
Given the usage model of the typical user described above, bandwidth consumption will be high during web page downloads or e-mail transmissions, and relatively low when the user is reading content of retrieved web page or downloaded e-mails. Figure 1 shows the standard bandwidth consumption of typical Internet usage. Since the majority of time online is not spent actually transferring IP packets, the average bandwidth consumption will be very close to the idle bandwidth consumption. In any case, it will fall well below 56 Kbps on average. Over time, we will see ‘peaks’ of bandwidth consumption, followed by a very low-level ‘idle’ consumption.
Figure 1 – Characteristics of typical Internet usage over time
Part B – Combining Broadband with Networking
The typical way to connect multiple users to a broadband connection within the same home or office is through a gateway/router. A gateway/router is defined as an access point to which the broadband link is connected on one side and multiple computers are connected on the other side. The gateway will typically have routing capabilities, allowing different packets to reach different computers. It may also include a firewall or other security measures to protect the computers from the ‘always–on’ Internet link.
What happens when more users are connected to the same Internet connection? Let’s consider the different types of users involved. Assuming that most Internet users will have the typical usage model (as illustrated in Fig. 1), the average Internet bandwidth consumption will equal the ‘idle’ bandwidth multiplied by the number of computers connected to the same link. Since the ‘peaks’ in bandwidth consumption are infrequent, the probability of different users’ peaks occurring at the same time is relatively low. Only when two or more computers attempt to use the maximum available bandwidth at the same time will they receive anything less than the maximum available bandwidth. On all other occasions, any user will feel as if he or she has the full bandwidth solely at their disposal. Figure 2 illustrates two computers sharing the same Internet access.
Figure 2 – Characteristics of a shared Internet connection
Given the typical bandwidth sharing illustrated in Figure 2, the logic behind sharing an Internet connection is clear. All users feel as if they have exclusive access to the broadband Internet pipe when, in fact, they are sharing. Having a Media Hog who continuously consumes high bandwidth from the Internet link might affect the sharing, but each user would still get half of the maximum available bandwidth.
Another benefit of sharing Internet access over a network is not having to install separate costly broadband modems for each computer and the cost savings of only subscribing to one broadband Internet account. When a $39.95 per month broadband access is shared among 4 users, the Internet access per user costs less than $10 per month, while each user feels that he or she is getting 100% of the bandwidth. These benefits can be considered to result from ‘economies of scale.’
Part C – Networking Technologies
Different networking technologies exist today for the short-range, gateway-to-computer link, also called ‘the last 10 feet’ (as opposed to ‘the last mile’ which describes the central office to gateway loop). These are based on different low-cost media that can be utilized within homes or small business environments, and reach up to 300 feet, which is typical coverage for such environments. Cahners In-Stat projects that in 2004, most home networks (20.1 million ports) will use wireless adapters. Phoneline networking will be close to 12.5 million ports, and power line networks will account for million ports. In small businesses, Ethernet (LAN) will probably remain the leading networking technology due to its higher attainable data rates.
Wireless networking uses FCC-approved unlicensed frequency bands, mostly in the 2.4Ghz and 5.4Ghz areas. Data rates can reach 54 Mbps, but wireless remains the costliest network to build and is highly susceptible to interference. The typical topology of a wireless network uses a wireless gateway, which is also referred to as an ‘access point’. The gateway is connected to the broadband Internet link and includes a radio transmitter and receiver to communicate with the computers and wireless cards plugged into the computers. Wireless cards include audio transmitters and receivers to communicate with the gateway as well.
Phone line networking (mostly through the HomePNA standard) uses the existing telephone line infrastructure (at frequencies not used by either voice or DSL communications) in order to communicate information at high data rates. Capable of transmission speeds up to 10 Mbps, HomePNA 2.0 is the fastest phoneline networking technology available today. The typical topology of phoneline networking includes a gateway connected only to the main telephone line for DSL communication, and simultaneous use of the higher frequency bands to communicate with computers. The computers themselves are equipped with HomePNA adapters connected to the telephone jacks in rooms where they are located.
Another medium that can be used to transmit information is the high-voltage power lines running throughout the home or office. Once again, high frequency signaling is used to transmit data between the gateway and the computers equipped with powerline adapters. One advantage of this technology is the fact that power lines that run into every piece of equipment can be used for networking as well. The disadvantage is that communication problems may occur because multiple high-voltage phases are distributed throughout the location and different computers may be connected to different phases that are not bridged.
Ethernet
Ethernet remains the most ubiquitous, least expensive, most reliable, and highest data rate networking technology. The only disadvantage to this technology is the need to install a dedicated wiring infrastructure based on Category 5 cables. While the cost of these cables is not very high, labor costs typically associated with such a retrofit are expensive, and can exceed $500 per port. When an organization moves into an office space that already has Cat 5 wiring installed, this is not a big concern. When a retrofit is necessary, however, the costs may lead to selection of other alternatives. These costs are typically referred to as ‘adds, moves and changes’. Costs like these drive businesses and individuals toward alternative networking technologies, and mostly towards wireless networking.
While wired networking technologies are more standardized, wireless technologies are heterogeneous. A variety of frequency bands, modulation techniques and networking protocols are used for the different standards, which make them incompatible with each other. Since wireless networking is in its infancy and deployment is not yet widespread, there is still time to decide on the winning technology. The different standards will continue to compete over this title.
802.11b
Currently, it seems as if this IEEE standard is picking up the most momentum as initiatives come into play such as Wireless Fidelity (Wi-Fi), an organization dedicated to ensuring interoperability between different 802.11b products. This standard is capable of delivering data rates up to 11 Mbps to distances up to 300 feet. 802.11b uses Direct Sequence Spread Spectrum technology (DSSS) in the 2.4 GHz frequency band.
802.11a
This newer IEEE standard, expected to be deployed in 2002, is relatively similar to the 802.11b standard. It operates in the 5.4 Ghz frequency band, and is capable of delivering speeds up to 54 Mbps. The range for this standard is lower, however, reaching approximately 100 feet. The technology barriers to implementing this standard are relatively high, which explains why no single company offers 802.11a-compliant products today. However, critics are questioning whether the traction gained by the 802.11b standard will create yet another entry barrier for new 802.11a products, which would be incompatible with the wide installed base of 802.11b products.
This initiative was founded to address the requirements of the home network, as opposed to the business network. Although sufficient for most residential applications to share a broadband Internet connection (averaging 1.5 Mbps or less), its 1.6 Mps initial data rate was relatively low. This standard uses Frequency Hopping Spread Spectrum (FHSS) technology in the 2.4 Ghz frequency band. HomeRF, Bluetooth, 802.11b, new 2.4 GHz cordless telephones and microwave ovens all share the same 2.4 GHz frequency band, which may cause a lot of interference. In fact, the most important value proposition that the 802.11a standard, which operates in the 5.4 Ghz frequency band, has to offer is the avoidance of that busy frequency band. The benefit is its higher data rate. Recently, an FCC ruling allowed the HomeRF standard to use a Wide-Band Frequency Hopping (WBFH) modulation, capable of delivering 10 Mbps. This makes the HomeRF a stronger competitor to the 802.11b standard. One major advantage that the HomeRF standard has over the 802.11b standard is price. With silicon prices dropping, however, this difference is shrinking to the point of becoming negligible.
Bluetooth was developed as a wire-replacement technology by several consumer product companies. Bluetooth has a very short range (a Type III Bluetooth device, transmitting only one milliwatt, can only reach 30 feet), with a maximum throughput of 721 Kbps. This is generally enough for a cellular phone to wireless headset implementation, which is probably the ‘killer application for Bluetooth. Due to the highly commoditized nature of the Bluetooth standard, the prices for Bluetooth devices are very low. It is popularly thought that in several years all cellular phones will have a Bluetooth module embedded in them. Although Bluetooth is not really considered a standard for wireless networking, it is included here due to its usage of the same frequency bands. Bluetooth falls into a category known as Personal Area Network (PAN) rather than a wireless LAN category. Some critics of Bluetooth claim that for a mere “wire-replacement” technology, its implementation is too costly, thus leading to the emergence of alternative low-cost short-range wireless technologies, such as the one developed by RFWaves.
Other wireless standards exist, for other geographical areas and other purposes. Some are proprietary. These standards include the HiperLAN /1, HiperLAN /2, RangeLAN, etc. As their adoption (especially in the US) is expected to be minimal, I did not discuss them in detail.
Part D – The Value Proposition of Wireless Networking
Using the traditional technology adoption curve to analyze the adoption phase of wireless networking technology, we see that this market is in its early adoption phase. It is used by more than the ‘innovators’ who bought a wireless network just because they had to have the latest technology. It is relatively easy to sell a wireless network to an early adopter, since he or she can appreciate the added value of wireless as opposed to wired. When it comes to the mainstream market, however, there are a number of factors that create value for the network buyer.
Data Rate
What is the desired data rate for a network? In answering this question it is important to appreciate the difference between actual need vs. perceived need. When comparing the wireless network to the wireless Ethernet system it will replace at rates up to 100 Mbps, the buyer is seeking to achieve the same data rate. Although sharing a 1.5 Mbps broadband link (whether through a DSL, cable, fixed wireless or even a T1 cable) would not take more than 1.5 Mbps, the buyer perceives a need for a higher data rate. This is one of the main reasons why 802.11b (11 Mbps) has gained more traction than HomeRF (1.6 Mbps) as the de facto wireless LAN standard. Although HomeRF now has the 10 Mbps bandwidth approval from the FCC that it sought, it still cannot catch up with 802.11b in the marketplace. One of the main reasons why the 802.11a standard (54 Mbps) is getting a lot of attention these days is its higher data rate. An excellent industry example will be the transition from 10 Mbps Ethernet to fast Ethernet (at 100 Mbps), and now to Gigabit Ethernet. Do we really need it? Probably not. Do we want it? You bet! Especially when the price difference is shrinking as technology advances.
Wireless networks raise the question of range between the gateway/access point and the computer. This range is also referred to as the ‘coverage area’. The gateway is typically installed in a central location, and all offices or rooms must be within the maximum range from the gateway in order to communicate. Wireless networks typically have ‘fall-back’ data rates. Typically, an 802.11b network will operate at 11 Mbps, but will drop down to 5.5 Mbps, or even to two or one Mbps, as a result of a range increase and insufficient signal-to-noise ratio (SNR) to support the more complex modulation techniques required for the higher data rates. The range that a wireless network must support should be comparable to its predecessor – the Ethernet. It should be noted that an Ethernet cable can be no more than 350 feet long. Installing a low-cost Ethernet hub as a repeater, however, can extend this range.
For some reason, users are more aware of security issues surrounding wireless technology than wired technologies. The standard 802.11b protocol implements Wired Equivalent Privacy (WEP), which is a 40-bit key encryption protocol, and implements the RC4-PRNG algorithm. Some 802.11b vendors implement a higher level of security using a 128-bit key. In order to be accepted, the level of security demonstrated by the wireless network must be equivalent to or better than that of a wired network Scientists from the University of California, Berkeley, claimed recently that the WEP encryption could be compromised. The response of the Wireless Ethernet Compatibility Alliance (WECA) is that the security level of the Wi-Fi products meets its objective of being at least as secure as the wired Ethernet.
Wireless networks are susceptible to interference by other radio frequency (RF) sources at the same frequencies. In the 2.4 Ghz frequency band, interference may be generated by other wireless networking devices using standards such as 802.11b, HomeRF or Bluetooth (a recent study was performed by Mobilian, covering the coexistence of 802.11b and Bluetooth networks). Moreover, interference in the 2.4 Ghz band is generated by new cordless telephones and microwave ovens. The other compelling reason to transition to the 802.11a standard (which operates in the 5 Ghz frequency band) is to avoid interference. Interference can also be generated within the network, by other nodes or by other objects that typically create interference such as multi-path fading to any wireless system (including metallic objects, walls, etc.). In order to replace a wired network, the wireless network must be immune to such interference at the same level.
A network must be fault-free, or at least as close to fault-free as possible. While computers running standard operating systems and applications may crash, a network gateway or access point must be fault resistant. The access point is typically installed in an inaccessible location. It should, therefore, be able to withstand errors (hardware or software) or other interference, and recover from them to provide durable and stable operation. This may warrant using different implementations than those used in the construction of a regular computer.
The operation of an Ethernet network became seamless by now. The ‘service discovery’ is automatic, and nothing has to be configured by the computer user in order to log on to the network (except user name and password, for security purposes). Currently, some wireless networking products require the user to be more technically savvy in order to use the network. Such differences must be eliminated, and the operation of the network must be completely seamless.
Easy Installation
While the installation of an Ethernet network is relatively simple and easy, the typical installation of a wireless network is significantly more complicated, requiring more time and technical depth. A premises ‘survey’ must first be conducted to find the locations for installation of the access point. Furthermore, a number of definitions must be programmed, most of which are directly related to the fact that the network is wireless. In order for wireless networks to become widely adopted, installation must not be more difficult than that of a wired network, if not actually simpler and automatic, thus creating a better value proposition for the wireless network.
One thing that designers and installers of wireless networks (or any other network, for that matter) must realize, is that the network may be, in some cases, the last component of an overall information system purchased by the organization (or household). Computers to be connected to the network already exist, and requiring the installation of a special PCI card may become a barrier to entry. Installing the network interface cards (NICs) should be a simple retrofit. When considering the interfaces that should be supported for installation of an NIC on an existing computer without opening it to install internal cards, there are several possibilities: a USB connection (probably USB 1.1 at 12 Mbps, has been on the market for several years), an Ethernet connection (through an RJ45, most enterprise/business computers have been shipped with built-in Ethernet cards for several years) or PCMCIA/PC cards for laptop computers.
When an organization or household chooses its first network (whether wired hub/router or wireless access point/gateway), it typically considers the existing number of computers. In the best case, it will consider the possibility of connecting a few more computers to the network in the future. As time progresses and computers must be added beyond the capability of the existing network equipment, however, that equipment often needs to be replaced with new equipment capable of supporting the new requirements. Scalability of network equipment is defined as the ability of the equipment to be expanded to support future needs. By definition, wireless networks should have the advantage in this area, simply because its access point/gateway only has one transmitter and one receiver to support a virtually unlimited number of computers. A wired network typically has a number of ports connected to it, and if more computers need to be connected to the network than ports available on the hub/router, then additional equipment must be acquired.
The residential market is probably the most price-sensitive. The small to medium business market comes in second. These markets will not tolerate high prices with high margins that enterprise/carrier networking equipment typically entails. Sometimes compromises must be made to lower the price in order to penetrate these markets. However, price is probably still the least important factor in purchasing networking equipment even in these markets, and will become the deciding factor once other requirements (order qualifiers) are met.
Conclusion
In conclusion, there are significant advantages of broadband Internet access if shared by several users and computers. Networks allow each user to feel as if he or she has exclusive access to the broadband pipe, while amortizing the costs of the broadband service over the total number of computers sharing it. It looks likely that the dominant networking technology will be wireless, and only time will tell which technology will rule that era. Nevertheless, in order for wireless networking to be widely deployed, it must still meet or exceed the value offered today by substitute networks, namely Ethernet.