Fixed wireless in an unlicensed band

In the 1990s, the Federal Communications Commission allocated 300 MHz of spectrum in the 5-GHz range, comprising the 5,150- to 5,300-MHz band and the 5,725- to 5,875-MHz band. These have since become the Unlicensed National Information Infrastructure (U-NII) bands.

Devices in the 5.725- to 5.825-GHz band operate with 1 W maximum transmitter output power and directional antennas of up to 23 dBi gain. The operating specifications avoid increasing interference with incumbent operators sharing the same spectrum.

There are three 100-MHz frequency bands in the 5-GHz band. The output power is limited to 23 dBm in the lower (indoor) band, 30 dBm in the mid range, and 36 dBm in the upper band.

The FCC set limits on emission levels outside the bands of operation and frequency stability requirements to protect adjacent spectrum occupants and sensitive operations that may operate on harmonic frequencies.

Transmitters operating between 5.15 and 5.36 GHz must be attenuated by a factor of at least 27 dB and within the frequency range outside these bands by a factor of at least 37 dB. In addition, all 5.25- to 5.35-GHz transmitter emissions within the frequency range to 10 MHz above or below the band edge must be attenuated by a factor of at least 34 dB. And for frequencies 10 MHz or greater above or below the band edge, they must be attenuated by a factor of at least 44 dB. All 5.725- to 5.825-GHz transmitter emissions from the band edge to 10 MHz above or below the band edge must be attenuated by a factor of at least 40 dB. Frequencies 10 MHz or greater above or below the band edge must be attenuated by a factor of at least 50 dB. These "rules" ensure the same level of interference protection outside all three bands.

A rigid channelization plan or mandate of a modulation efficiency standard has not been defined by the FCC because it did not want to delay implementation of U-NII devices by precluding certain technologies or applications. The low power limits will ensure efficient use of the spectrum by providing for high-frequency reuse, which will allow for large numbers of U-NII devices to share the spectrum in any geographic area.

However, the type of devices that will be approved for this band are defined. FCC regulations Part 15 says operations in the 5.15- to 5.35-GHz and 5.725- to 5.825-GHz bands will be limited to wide-bandwidth, high- data-rate digital operations. This will give equipment manufacturers the flexibility to design and manufacture a variety of broadband devices using different technologies and modulation techniques, while ensuring that this spectrum is used for its intended purpose. This definition will be enforced through the Commission's equipment certification process.

The U-NII band is divided into 15-Mhz channels with guard bands for frequency spacing. Frequency spacing refers to the actual bandwidth space that is allocated for every channel out of the total spectrum.

No standards require specific channel pairing use. Vendors are free to choose the cellular model using two channels (one for upstream and one for downstream traffic) or a single channel. The mechanism used to support the upstream and downstream traffic will determine the end-user concentration within a cell.

The customer-premise "radio"--typically known as the subscriber unit (SU)-contains the antenna having a 20 degrees by 20 degrees directional beam, radio transceiver, a modulating modem, media-access control (MAC), and a switch or router contained in a housing mounted on the side of the building. The connecting interface is a 10 BaseT, 100 BaseT or ATM25 going to an indoor junction box with a RJ-45 interface to the customer's equipment.

The wireless basestation or cell is composed of sectors. The basestation is the master unit supporting a number of SUs. Physically it is identical to the SU except that each access-point antenna forms a 60 degrees by 7 degrees pattern along the horizon. The connection to the Internet "cloud" is typically a DS3 or OC3.

Making the link

At the customer premises, an Ethernet hub, ATM switch or a PC equipped with an Ethernet or ATM card is plugged into the indoor junction box where the SU will scan for the best access point (AP) to lock into, and automatically register and provide an Internet address for the PC, if needed. Depending on the protocol configuration selected, the Ethernet data, for example, is either encapsulated into ATM cells or the Ethernet layer is stripped, leaving Internet Protocol packets that are segmented into ATM cells.

Basically, the SU sends the data to the AP over the channel using either time-division duplex (TDD) or frequency-division duplex (FDD). In Adaptive Broadband's case, both upstream and downstream traffic time-share a single channel. To transmit broadband traffic (such as voice, video and data) various vendors may support AAL1 or AAL2 and AAL5 type cells.

Licensed spectrum offers the promise of regulated security, thereby reducing risk to the client. However, unlicensed broadband wireless vendors can offer standard encryption techniques as well as authentication at the MAC layer and supported protocol, scrambling and physical security.

Basically, the FCC uses competitive bidding and auctions to grant licenses for spectrum. This mechanism is quite lucrative (tens of billions of dollars) for the U.S. Treasury. But the expense and time surrounding getting the grant has pushed the smaller service providers to the sideline. However, the unlicensed spectrum enables smaller service providers as well as private networks to quickly acquire a broadband fixed wireless solution.

There is a wide range of consumer/business applications that use the wireless spectrum, such as TV, medical facilities, mobile phones, high-speed Internet connectivity, and broadband voice, video and data. In order to move fixed broadband wireless access to the masses, various coding schemes are available, such as VOFDM (vector orthogonal frequency-division multiplexing), OFDM (orthogonal frequency-division multiplexing), time-division multiple access, code-division multiple access and Docsis (data over cable service interface specification). Some solutions are combining VOFDM and Docsis, but many in the fixed wireless industry believe fixed or mobile radio is a sufficiently different transport medium from coax and fiber that it requires an unencumbered approach.

When building a U-NII solution, several essential questions need to be considered by the designer:

• Exactly how do you grant bandwidth to a user (packet-on-demand, channel-on-demand or dedicated)?

• How do you divide the radio channel to support upstream and downstream traffic (TDD vs. FDD)?

• Exactly what is the right modulation technique to put onto the radio carrier in order to get the best efficiency (VOFDM, TDMA, CDMA, etc. and QPSK, 16QAM, 64 QAM, etc.)?

• And, what service levels do you provide (for example, asynchronous ATM levels of service, proprietary bandwidth-on-demand, and Docsis)?

The overall goal is to build a solution that provides the highest consistent service possible in a given bandwidth over the greatest geography reliably.

The consortium backing VOFDM as a coding standard pairs Docsis (leveraging the success of the cable modem industry) as the MAC layer. VOFDM would specify coding, while Docsis would detail bandwidth access and service definition. However, the efficiency of Docsis in a wireless environment is an ongoing argument. This is due to the fact that wireless devices require frequency tolerance and tracking. The bottom line is that wireless transmission is significantly different from cable transmission--it requires a unique MAC to achieve efficiency and reliability in a fixed environment.

In order to address the highest consistent bandwidth allocation, a developer or user can allocate bandwidth-on-demand where each user is assigned a fixed bandwidth or packet-on-demand where a user's packets are assigned bandwidth when required.

Both FDD and TDD support dynamic balance between upstream and downstream traffic. Some engineers argue that time-division duplex is 3 dB worse than frequency-division duplex in terms of link budget. Given equal bandwidth channels of 1 MHz and a modulation scheme, which achieves 'b' bits/Hz/second, FDD, in full duplex mode, achieves 'b' Mbits/s downlink and 'b' Mbits/s uplink. While TDD achieves 'b'/2 Mbits/s uplink and 'b'/2 Mbits/s downlink (a total throughput of 'b' Mbits/s). In order for TDD to have the same throughput, it needs twice the bandwidth and therefore suffers 3 dB more noise (its link budget is 3 dB worse). However, FDD systems use two channels (2 MHz) and TDD systems use only one channel (1 MHz). For a given channel bandwidth, TDD link budgets are no different from FDD link budgets.

In terms of a throughput comparison, the spectral efficiencies are identical for FDD and TDD, for any given bandwidth and fixed uplink/downlink load share, if both systems use the same basic modulation scheme. However, even though TDD only transmits half duplex, FDD requires twice as many frequencies.

TDD can vary the transmit and receive temporal split and therefore match any transmit and receive load split. Meanwhile FDD, on the other hand, is set up by the service provider and has no dynamic flexibility. And, the problem is that this makes FDD extremely inefficient, except when the traffic load exactly matches its fixed capability.

Essentially, all radios that are being built today have limited adjacent-channel rejection when the channels are contiguous.

The use of guard bands may be indicated in both FDD and TDD methods to cut down on adjacent-channel rejection.

With cellular reuse configurations and high-gain antennas, FDD can operate without the need for a transmit/receive (Tx/Rx) guard band, by using physical isolation. However, FDD systems do Tx and Rx simultaneously, and require large isolations between Tx and Rx frequencies (around 120 dB). If multiple FDD operators use the band and access points are not co-located, to reduce system cost, they will be separated by a guard band, enabling all terminals to use the same filter (diplexor).

Because TDD systems do not transmit and receive simultaneously, they do not require Tx/Rx guard bands. But two TDD systems in adjacent bands can mutually interfere unless the APs are co-located, which is usually not feasible. The solution is a guard band to prevent interference.

TDD pluses

A TDD system and FDD system may reside in the same band. There are interference possibilities, but they are no worse than two FDD systems. In both cases a guard band and filtering are necessary to allow coexistence without interference.

Time-division duplex has several benefits over frequency-division duplex:

• It allows dynamic allocation of uplink /downlink bandwidth.

• It does not require paired frequency allocation--it's much more flexible in terms of frequency allocation.

• It has simpler hardware--that is, no diplexor.

• Spatial diversity can be implemented at the basestation only, reducing CPE cost.

• Channel equalization can be performed at base station only, reducing equipment cost.

• Adaptive channel equalization combined with Tx predistort can improve resistance to multipath.

• TDD enables the use of simpler and more effective adaptive antennas.

• It has more effective power control.

• And, it is easier to self-test--by closing Tx/Rx loop at each modem.

A successful solution mandates service level guarantees. Leading fixed wireless devices have implemented ATM from the wireline world because of its ability to define and guarantee a nearly infinite number of service combinations. In this way U-NII solutions can easily interoperate with terrestrial ATM networks.

See related chart

See related chart