Wireless communications networks

A wireless communications network is any system which uses electromagnetic waves to communicate wirelessly over some distance. Common applications are cellular phones, CBs, Ham radios, wireless local area networks, point to point links, sensor networks, and personal area networks. Distances range from several feet to tens of miles.

Here we give an overview of different technologies, critical applications, and the OSE projects to address them.

Tools, projects, and components

OSE

• Point To Peer
• Communications tower
• Low-cost, off-the-shelf, 802.11 based broadband deployment for backhaul and last mile applications

OSI model

This article will be roughly structured to follow the OSI model[1], and it will cover the first 4 layers.

Physical details

This section deals with details applicable to all wireless systems, constrained by physics, economics, and ubiquitous use.

Frequencies

Effect on antenna

In order to get information from a circuit board into the air, it must pass through an antenna. Antennas only work at certain frequencies - the higher the frequency, the smaller the antenna can be. Microwave frequencies allow antennas to be small enough to mount onto an access point, small satellite dish, or on a tower. Frequencies lower than microwave frequencies can be used, but the antennas must be larger.

There is a trade off between antenna size and gain. The larger the antenna, the higher the gain, if frequency is held constant. If the antenna size is held constant, the higher the frequency, the higher the gain. If both frequency and gain are specified, the antenna size is then dictated. Antenna size and frequency really are proportional. Antennas can be either smaller than an inch, or miles, depending on the corresponding frequency.

Effect on propagation

Besides determining details of the size of the antenna, different frequencies propagate differently. In general, lower frequencies penetrate buildings, foliage, and other obstacles more easily. Very low frequencies, such as used by some ham radios, even bounce off the ionosphere, and can reach around the globe. High frequencies can be absorbed more easily by obstacles. Because higher frequencies can be more easily directed because they need smaller antennas, they are usually more directional too.

Legal regulations

Frequency use is extremely regulated. Because of this, only certain frequencies can be used for certain purposes. One important thing to remember is that even though low frequencies can penetrate well, there is less bandwidth available (there is only 500 MHz available from 0 - 500 MHz, however from 5 - 10 GHz there is 5 GHz available). For all intensive purposes, we are interested in the ISM and UNII bands - 900 MHz, 2.4 - 2.5 GHz, and 5 - 5.8 GHz. In the near future, 60 GHz and 700 MHz may be available to the public (technically they are available, but the hardware isn't readily available, cheap, or ubiquitous).

Those with an amateur radio license enjoy the use of frequencies not available to the public, and are also allowed to operate their own hardware without it being certified by the FCC beforehand. Amateur radio licenses are relatively easy to get compared to the amount of freedom they give.

Range

Generally, in a line of sight situation, higher frequencies can create longer links, due to increased gains in the antennas. For a ubiquitous signal that reaches everywhere, a low frequency signal will have farther range because it goes through buildings and foliage more easily.

Antennas

There are various types of antennas for various applications. At microwave frequencies, there are two main types: directional and omni directional. Directional is either used in point-to-point links, or in point-to-multipoint links (on the client side). Omni directional antennas are used in general devices such as laptops and cell phones, and specifically for access points trying to cover a broad area, for example in a living room or on a tower serving a valley.

Common directional antennas are yagis, parabolic grids, and parabolic dishes. Common omni directional antennas are dipoles and sectorized antennas.

Generally linear polarization is used except in satellite systems.

High-level transceiver architecture

A transceiver is a transmitter + a receiver.

A transmitter takes a modulated baseband signal and mixes it with an RF carrier (continuous sine wave or CW), using a mixer. The baseband can either be an analog signal or a digital signal. In the case of a digital signal, generally bits encoded with forward error correction are transformed into the frequency domain via an FFT operation carried out in dedicated ASIC hardware (or in the case of a soft radio, just dsp). Then DACs convert to a signal which is mixed.

Receivers work the same way but in reverse. First the tiny received signal is run through a filter to get rid of adjacent channel noise. The received signal is down-converted with a local RF carrier (the frequency of this carrier is the "tuning"). Then ADCs get the digital bits, and it is decoded.

Generally, the same baseband signal can be unconverted or down converted to any frequency.

Link budget

To determine how far two radios can communicate with each other, whats done is called a link budget calculation. This calculation takes all losses and gains, in dB, and subtracts and adds them to get the net result. There are plenty of link budget calculators that can be used, including the Ligowave calculator which downloads terrain data to find obstacles.

Hardware cost

RF hardware is expensive. It must be geometrically very accurate, and the materials used must be pristine. Luckily, highly refined silicon manufacturing techniques has dramatically reduced the cost, and made wireless a reality in the home. This applies only to commoditized wireless hardware, such as 802.11, Bluetooth, and cell phones, that implement mass produced RFICs. Specialized RF hardware, such as highly sensitive, high-power, or using uncommon licensed frequencies is still very expensive. Lower frequency hardware is generally cheaper though, such as ham radios or CBs.

Today, in quantity, 802.11 modules are less than $10, and Bluetooth modules are less than $5. Of course, when integrated into a product, the cost will be higher - additional circuitry, driver software, regulatory and industry certification, and marketing will all add to the cost.

Wireless communications technologies

Modulations

There are quite a few different modulations, as described by Wikipedia[2]. For digital communications, the two most important are spread spectrum and ofdm. Spread spectrum multiplies the original signal by a pseudo random code resembling noise. The code is much larger than the information it is carrying. On the receiving side, the same code is used and correlated to the received signal. The correlation function has an associated "gain" that takes a signal that seems to be in the noise brings it up to a higher signal to noise ratio. 802.11b uses spread spectrum (dsss). One benefit is that if the code used to receive is different than the code used to transmit, the signal will look more like noise. So if there are multiple transmitters and receivers on the same channel, it may be possible to communicate regardless.

One problem with spread spectrum is it is more susceptible to fading and multipath.

OFDM (orthogonal frequency division multiplexing) is a more recent technology than spread spectrum. It uses many narrow subcarriers/channels within the operating channel to transmit many parallel data streams. The benefit of this is high spectral efficiency and robustness against fading and other problems with the channel. The reason it is more robust is because it is composed of many slow streams instead of one very fast stream. Also, if a piece of the channel has interference, some data will still get through. One challenge of ofdm is it requires amplifiers which are linear over a larger range, because of the high peak to average ratio of the signal. It is also more dsp intensive, though nowadays that usually isn't an issue.

Most new wireless technologies use ofdm - 802.11n, LTE, and WiMax. It is generally viewed as the best.

Media access

Media access refers to how the communications medium is accessed - in this case the air (as opposed to copper or fiber). There are special circumstances for an air medium, such as hidden nodes.

There are a number of medium access technologies, and it is important to understand them.

CSMA/CA

Carrier Sense Multiple Access/Collision Avoidance - This is what 802.11 uses, and is similar to how ethernet works. Essentially, the receiver listens to the channel. If the channel is in use, it waits. Once the channel is clear, it waits for a random time (within a specified min/max range) and if the channel is still clear, transmits. If another client transmits within the countdown period, it resets the timer to a new random value. This is efficient for small networks of heterogeneous devices because it uses available airtime and doesn't need a central scheduler or master.

As the number of clients grow however, the model starts to break down and there are too many collisions.

TDMA

Time Division Multiple Access - Divides time up into many different time slots, and then allocates time slots to clients as needed. Precise timing is required when a large network like a cellular network uses this method. GSM is a TDMA technology. It has tighter control over bandwidth allocation and avoids collisions with large numbers of clients.

CDMA

Code Division Multiple Access - Differentiates clients by each client using an orthogonal code. In other words, this is spread spectrum with each pseudo random code being orthogonal. When the correlation function is run, orthogonal codes have no interference or affect on the ability to receive the desired code, as long as all the received signal levels are relatively the same between clients. CDMA cell networks use this.

OFDMA

Orthogonal Frequency Division Multiple Access - Divides the entire band into orthogonal subcarriers, which can be assigned to specific clients. In other words, the 'parallel subcarriers' mentioned above now cover not only a single channel, but the entire band. And not just one device uses them at a time, but different devices are assigned subcarriers. LTE and WiMax use this.

MIMO

MIMO (multiple input, multiple output) is a mathematical technology which codes signals in a more complex way inside of the dsp, and feeds the digital outputs to multiple baseband tx chains. These baseband signals are transmitted simultaneously and received on multiple rx chains on the other side. The digital hardware then takes these multiple received bits and combines them. Using this special coding, it is possible to multiply the bandwidth by the number of spacial paths taken by the signal. In a building, there may be many paths, and therefore the bandwidth can be multiplied by 2 or 3 times (alternatively, if the bandwidth is the same then the entire message takes less air time, and a denser network can be used. alternately, if the first two are held constant, a more reliable network can be used).

For point to point links, dual polarized, high-gain antennas are used to attempt to get two separate data streams in both available polarizations.

MIMO is strictly a digital technology because of the math involved on bits.

It should be noted that generally regulations specify that each separate tx chain has an output power such that the entire device does not exceed regulations as a whole - for example, a single transmitter may be able to transmit 100 mW, but a 2x2 mimo unit can transmit a maximum of 50 mW on each transmitter.

Another thing to note is that MIMO can still be beneficial when the other side (transmitter or receiver) is a legacy device or has a single antenna. The benefit is beamforming on the tx side, and maximal ratio combining on the rx side. Note not all radios implement all aspects of mimo.

Protocols

Wireless protocols can be complicated. Generally, the firmware on a chipset is an extremely highly guarded piece of code. The firmware in combination with the driver implement the wireless protocol. Much of the complexity comes from all the features that wireless protocols contain, including roaming, power save, qos, and edge cases to prevent poor performance. Another factor is that different manufacturers must test their wireless devices with devices of other manufacturers, to make sure they all actually implemented the standard in a compatible way.

In general, if one wanted to make a completely open source radio design, the hardware portion would probably be pretty easy, but the software portion would be difficult.

Bluetooth low energy might be an exception here - there are only 5 states in the state diagram.

Mesh networks

It must be noted here that the only thing separating normal wifi or other devices from forming mesh networks is the lack of a protocol (one that is widely implemented anyway). The hardware is cheap and available, it is the mesh protocol. Currently, an IEEE task force is working on a mesh networking standard - 802.11s.

There are other mesh network protocols in existence such as Zigbee. More at about Mesh Networks at Point To Peer