RF Analysis

Once the site surveyor has been able to establish whether an airlink is obstructed, and how
severely, if obstructions exist, the time has come to evaluate the RF environment for that airlink
and determine whether it is ultimately useable. This happens with either a dedicated spectrum
analyzer or a specialized software program run over the actual data radio equipment.
All the major instrumentation companies—including HP/Agilent, Anritsu, Wiltek, and
Rohde & Schwartz—make dedicated spectrum analyzers, and most make handheld models
that can be managed by an individual obliged to mount a radio tower. These are precision
instruments capable of wideband tuning and of detecting activities in nearby bands that may
impact transmissions in the bands selected by the network operator. While not inexpensive,
they are designed for outdoor use in adverse weather conditions and will serve the operator
well over years of installation work. And if operators are unwilling to purchase such a device,
they can rent them, though I strongly recommend purchasing one since in a growing network
it will be in more or less continuous use. Also, if the adaptive array antennas are used, a special
type of spectrum analyzer known as a deep memory waveform generator will be required.
Agilent is the principal manufacturer of such devices. Increasingly, broadband radios include built-in spectrum analyzers, reducing the need for
specialized instruments. In all cases the radio itself will be far less portable, however.
Unlike the determination of line of sight, the RF examination should be done at both ends
of the link, since there is no way of determining the presence of interference away from the
point where it is experienced. Make sure that readings are taken at the precise location where
the antenna is mounted. It does little good to make measurements from street level if the
antenna is to reside on the roof of a six-story building. It is also advisable to take a number of
measurements over the course of several days since interference is often intermittent. As a network
operator, you will be obliged to offer high performance over your network at all times, so
you must be aware of any potential problem.
In effect, each individual link requires its own site survey, though network operators
striving to launch a network will most likely confine themselves to a round of surveys encompassing
the most promising customer sites. It is also a good idea to get a general reading of the
entire metropolitan area where one has chosen to situate one’s network through random spot
surveys. If, for instance, one finds that the city center is already densely occupied with private
WLANs utilizing 2.4GHz, then one may want to consider another unlicensed band. In performing
such a general reading, it is a good idea to rent a crane or a vehicle equipped with a
hydraulic “cherry picker” so that readings can be taken at heights where an antenna may ordinarily
be installed.
The purpose of the spectral analysis phase of the survey is simply to measure the level of
interference at the antenna site and nothing else. The spectral analysis does not and cannot tell
one what the signal strength at the antenna will be.
At the point when network operators have completed a spectral analysis of the two ends of
the airlink in question, they must then come to some conclusions about the integrity of that
airlink and its suitability for providing networking services. Such a determination involves
some rather complex technical issues.
The mere presence of interference does not rule out using the link for data communications.
All commercial radios tolerate some degree of interference, and, in the case of radios
operating in unlicensed band, the presence of significant background interference is assumed.
Interference becomes critical when the level of interference approaches that of the signal.
Individual radio designs differ in their ability to reject interference within the same channel
and adjacent channels as do different modulation techniques in general. For instance,
frequency hopping confers an especially high immunity to interference because the signal
becomes in effect a moving target as it hops from one channel to another in a synchronized
pattern. Coarse modulation techniques such as 16 quadrature amplitude modulation (QUAM)
also provide superior rejection of noise and interference. Unfortunately, however, those techniques
that render the airlink more robust in the presence of interference may also limit the
speed of the connection or reduce the electrical efficiency of the radio.
The issue is further complicated by the fact that the TCP transport protocol, mandated in
the case of 802.16, has provisions for resending packets in the event of interference-induced
losses. So long as the interference has some degree of intermittency—that is, it falls below the
level of the signal sufficiently often to permit some packets to go through—a transmission can
be maintained even in the face of heavy interference, though at considerable cost of speed and
reduction of service quality. The question then becomes, what is acceptable to the customer?
One should also understand that the implications of network nodes operating in a highly
interference prone environment go beyond the individual customer. To the extent that an individual
radio is doing a lot of retransmitting, that radio itself is elevating the general noise floor in the area and causing other radios to retransmit more frequently, which in turn further
degrades the performance of the first radio in a vicious circle. In a sense, radio interference is a
bit like acoustic feedback in a public address system. One can get only so much level out of the
system before escalating feedback buries the signal.
The equipment manufacturer that one has chosen to work with should be able to provide
information about levels of the interference in terms of decibels below the reference level of
the signal that the radio can tolerate while maintaining a good data link with few resends. If
interference regularly exceeds the recommended level, then one must consider abandoning
that link.
The temporary installation of radios at both ends of the link will provide the final determination
as to whether the link is useable. The radio’s own diagnostic software should permit the
surveyor to measure signal strength and to evaluate the availability of the link. At this point in
the process experimentation with the placement of the antenna is entirely in order, and the
operator may find that adjustment on the scale of centimeters can make the difference
between a good airlink and its opposite.