Training: RF for Wireless Surveillance

Wireless surveillance failures and problems are well known. These deployments are especially hard because they demand expertise in surveillance, IT and RF. Have problems in any one of these areas and the likelihood that your system fails increase greatly.

In our experience, out of these 3 areas, security integrators typically have the least amount of expertise in RF. For 'normal' wired surveillance projects, you can be an excellent integrator only knowing surveillance and IT. However, for wireless, understanding how video is transmitted by radio is critical to avoiding big mistakes. 

Making this more challenging, using RF in surveillance forces lots of unclear tradeoffs:

• The more power your radio transmits, the more likely your video will 'make it' to the other side. However, you need to be cognizant of legal limitations in the commonly used unlicensed frequences.
• The narrower the beam width of your antenna, the further your camera can be from your site. However, this can make it more difficult to line up your radios and can cause problems in designing systems that 'talk' to multiple cameras.
• Unlike wired transmission which is generally very stable, wireless surveillance throughput can vary significant, can drop out of the blue or due to the weather or vegetation growth. Integrators need to factor in potential issues and plan for likely risks.
• You can choose from many radio frequencies but you need to be careful because important tradeoffs exist in bandwidth capacity, interference likelihood and ability to transmit through obstacles.

To this end, unlike wired, you generally need to carefully plan wireless systems and run the numbers to make sure you can accomplish what you propose. This is even more complicated because you have to be explicit and factor in your resolution, your transmission control (e.g., CBR vs VBR), the scene complexity captured, your frame rate, etc. Even if you establish a link, the link may be insufficient to deliver the bandwidth you need (or implicitly promised to the user).

In this report, we provide videos and explanations of the basic concepts and key issues in using radios to transmit surveillance. Our goal is to help you understand the fundamentals so you can avoid mistakes and know where to focus your efforts in designing and deploying projects.

The first half of the report is a written tutorial/analysis of wireless fundamentals. The second half are 2 videos that show these concepts in action. 

You can read through in order or jump ahead to the videos at the end.

Impact of Distance on Signal / FSPL

As a signal is propagated from transmitter to receiver it has a tendency to lose power along the way. This phenomenon is known as free space path loss (FSPL). 

Key points about free space path loss :

• Higher frequencies will lose more signal strength than lower frequencies over the same distance
• It is an exponential loss, proportional to the square of distance * frequency

For example, 2.4GHz radios generally provide longer ranges or larger coverage cells than 5.8GHz radios. A 900MHz signal will experience even less signal loss over a given distance than either 2.4GHz or 5.8GHz signals.

Keep in mind the trade-offs to these properties. A 2.4GHz signal, while generally can cover a longer distance than 5.8GHz, can not 'pack' as much data into the signal as 5.8GHz. Also, 2.4GHz spectrum is widely deployed and offers fewer channels. This makes 2.4GHz more susceptible to interference issues. The 5.8GHz spectrum carves out many more channels than 2.4GHz and is less widely deployed at the time.

Also consider that the typical UDP transport method of video makes it more susceptible to packet drops caused by RF interference. Moreover, temporal video compression schemes such as H.264 are highly intolerant to losses of I-frames. (A loss of I-frame means P- and B- frames are not decodable as well). Therefore 5.8GHz radios because of their channel diversity and lower deployment are preferable to 2.4GHz radios in video applications.

Impact of Obstructions on Signal

Free space is not the only thing in the environment that attenuates (or decreases) signal strength. Obstructions have various negative impacts to signal propagation, as well. 

Let's highlight these effects/impacts:

• Absorption
• Reflection
• Multipath

Obstructions generally absorb and reflect radio signals, and thus reduce the level of signal reaching the receiving end. The level of attenuation is dependent on the composition and reflective properties of the material. For example, elevator cabs are surrounded by dense concrete and metals, resulting in large degrees of signal loss. On the other hand, dry-wall is significantly less absorptive/reflective.

Multipath is a type of reflection, whereby, the signal eventually finds a way to the receiving end. However, the signal will be out of phase (or sync) with the 'primary' signal. These effects can be unexpectedly adverse to a link's quality. Environments with a high degree of reflective surfaces  - e.g., water, glass, mirrors, foliage etc. - are prone to multipath signal loss even in shorter range or WLAN applications.

Impact of Antenna Selection

Many people are familiar with the 'rubber-duckie' antennas seen on home wireless routers. These antennas are formally known as dipole antennas. They are essentially a type of omni-directional antenna that propogates the signal 360 degrees perpendicular to the antenna's axis. This is a very appropriate antenna selection for home use because of its ability to propagate throughout a wide area - e.g., bathroom, kitchen, bedroom etc. Also, little to no expertise is required to align the antennas.

However, a dipole antenna of this type will not be appropriate for other wireless topologies. For example, a point-to-point (PtP) link between two branch offices several miles away would be better served with a more highly directional antenna - such as a high gain parabolic.

Some antenna types (least directional to most directional):

• Omni-directional - e.g. 6dBi
• Patch/Panel - e.g. 12dBi
• Sector - e.g. 18dBi
• Parabolic e.g. 21dBi

The key concept to understand about the various antennas is that they do not provide additional power to a radio signal. They basically 'concentrate' the power of the signal. This level of 'concentration' is known as gain and is quantified using a unit called dBi. The higher the gain the greater the concentration into one direction and thus the further the signal can be propagated.

Antenna manufacturers usually provide beam pattern diagrams for their antennas. Beam pattern diagrams show the 'shape' of the signal propagation. Beam patterns are often reduced to angles for the sake of simplicity and classification - these angles provide a good starting point for antenna selection. For example, while an omni-directional antenna by definition has a 360 degree beam-pattern, a parabolic, which is highly directional, may have only a 5 degree beam-pattern.

It's easy to see the trade-offs between the various antenna types. Low-gain omni-directional antennas can cover front, back, and side areas, and do not require the compelxity of antenna alignment, but they offer shorter distances. The middle ground patch/panel and sector antennas provide better distance but narrow the beam pattern; many times they can be 'eye-balled' to align sufficiently. Parabolics are optimal for long distance PtP links, but they are the most problematic to install and align. Also they are susceptible to sway, wind, vibration etc. as even a small movement of the underlying base can throw the antenna off alignment.

In high bandwidth video applications it is generally advisable to opt for the higher gain antennas rather than the lower gain ones. The reason is that the data rates of radio links are highly dependent on the signal strength level. For example, five HD cameras might consume 15 mbps which would essentially require the full data rate (54mbps aggregate or 27mbps one-way) in a 802.11a system. Thus, the higher gain antenna will help ensure adequate signal levels for reliable transmission.

Signal to Noise Ratio (SNR)

Most wireless products display an SNR value to help indicate the quality of the link. The value is determined by two key factors:

• Received Signal Strength
• Noise Floor

For example, if our noise floor is -90dBm and our signal is a strong -60dBm then our SNR is 30dB.

Note that many radio systems employ adaptive rate shifting, whereby the bandwidth of the link is determined by the received signal strength or SNR. Thus, for high bandwidth video applications, it is imperative to maximize the SNR or received signal strength.

Link Budgeting

A link budget is a calculation aimed at estimating the signal level that is received at the other end of the link. It is an important tool in the design phase when specifying radio systems such as transmit power, frequency, antenna type, cables, connectors and so forth.

The end goal of the calculation is to design the right radio system to achieve a desired signal strength. This signal strength will determine the data rate and reliability of the link. Thus, in video applications it is important to

Link Budget Positives:

• Transmit Power - Power in dBm or mW
• Antenna Gain (Tx) - Gain in dBi of the transmit antenna
• Antenna Gain (Rx) - Gain in dBi of the receive antenna

Link Budget Negatives:

• Free Space Path Loss - Calculate using equation with distance (km) and freq (MHz) as variables
• Cable Loss - dB loss per foot or meter of RF cable length
• Connector Loss - dB loss for RF connectors/splitters etc.

Let's calculate an example where our radio is transmitting at 16dBm (30mW) using a 8dBi patch antenna at both ends of the link. This yields 16 + 8 + 8 = 32dBm of signal power.

We'll estimate our cable and connector losses at 2dB. The 'big' loss is going to come from our free space path loss:

FSPL(dB) = 20Log(d) + 20Log(f) + 32.45

The units for d (distance) is km and MHz for f (frequency). All logs are base 10.

Our example will use 1km and 5800MHz. After plugging in the values we get ~108dB of FSPL.

With our connector/cable losses added in we get 110dB of total loss.

Our final budget calculation gives us 32 - 110 = -78dBm

Note that the -78dBm signal is fairly weak for video applications and doesn't provide much of a fade margin (signal fluctuations due to environmental factors). Performance could be increased significantly with parabolic antennas. For example, using 20dBi parabolics at both ends would yield a receive level of -54dBm instead of -78dBm.

In general most radios require receive signals in the lower -70s (e.g. -73dBm) to provide the highest data rates. Generally, the goal for video applications should be to acheive the highest data rate possible. While -78dBm is close to the low -70s it does not provide much margin. A good fade margin would be 15dBm. Thus, if our receive level for maximum data rate is -73dBm, we should target roughly -58dBm in our link budget).

Connector and cable losses are usually quite small compared to other radio factors (e.g. fspl, antenna and transmit powers) but designers should be aware of some strategies to minimize their affects. For example, radios with integrated antennas virtually eliminate losses due to RF cable runs. In the 'old days' separate radio and antenna units had the advantage of not requiring power at the top of poles/towers. But with the advent of PoE powered radios this is becoming less of an issue. Also video applications require power (e.g., PoE to cameras) to the external poles/towers anyways, negating one of the major advantages of separate radio and antenna units.

Professional wireless network design tools (e.g., EDX) take into account geographically specific information. Factors such as rainfall, fog, atmosphere, terrain and other obstructions etc., make the calculation much more sophisticated for business critical telecommunications systems. However, pricing and knowledge requirements generally preclude their use for security integrators.

As a compromise consider one of the many online path loss/link budget calculators. They do not have the benefit of geographical specificity, but if the proposed link has clear line of sight (LoS) these calculators will give a fairly reliable estimate of link quality. See this online tool as an example.

Bandwidth Potential (Frequency Band, Channel Size, Modulation)

From a pure physics standpoint the potential data-rate is based on channel width. For example, both 2.4GHz and 5.8GHz 802.11 (a,b,g) radios use a channel width of 20MHz. This makes their maximum bandwidth potentials the same. For example, if using the same modulation scheme the 20MHz channels from either the 2.4GHz or 5.8GHz radio will exhibit the same data-rates.

The difference between 2.4GHz and 5.8GHz frequency bands is really the total amount of available spectrum in which to carve out multiple 20MHz channels. The 2.4GHz ISM band has only about 100MHz of spectrum. The 5.8GHz ISM band has roughly 100MHz as well, but has a few more UNI-I bands to work with as well. That is why there are more non-overlapping 20MHz channels available in 5.8GHz radios implementing both ISM and UNI-I bands.

If channel width is the 'raw' muscle for wireless bandwidth, then modulation is the 'brains' or logic behind it. For any given channel width, the efficiency of modulation will determine the bandwidth capacity of the channel. There are many modulation types used in 802.11 wireless networking. Some of the common ones (in order of least to most effective):

• BPSK
• QPSK
• CCK
• OFDM (BPSK, QPSK, QAM-16, QAM-64)

For example, CCK, is the modulation scheme that allowed 802.11b to provide 5.5 and 11 mbps of aggregate bandwidth on the same 20MHz channel versus the 2mbps maximum data rate previously achievable using DQPSK.

Note that OFDM has various modulation types available for it's sub-carriers. For example, OFDM/QAM64 is used to achieve the 54mbps data rate in 802.11a/g radios.

Video Training - Configuring RF for Surveillance

Video Training - Using a Link Budget Calculator