Useful Tools

contact us contact tutor/group email to a friend accessibility options report a fault
Communications IC Architecture

Communications IC Architectures

Unit 7: Wireless WANs

 

Unit Contents

7.1 Applications of Radio - Wireless Local Area Networks

  In this chapter we will ask, and answer, several important questions in the radio applications area. The nominal topic will be wireless local area networks, but the conclusions will have surprisingly wide applicability, especially in the burgeoning field of short range, low power radio.

The first questions to ask are:-
 


And, slightly upstaging the rest of the chapter, the answer: -  Mobility is the Key


[back to top]


7.2 Mobility is the Key

What are WLANs?

Do not forget the simplicity and freedom from interference which cables can give! Mixed cable and wireless systems may be the best option.  
Topics to be covered in this section:-
 

 

 

Fig 1: Simple WLAN

Simple WLAN

 

The diagram above illustrates a simple wireless LAN, connecting several PCs with each other and a printer. At present, this is usually achieved by a mess of wires around the back of the PCs and trailing across, or more expensively underneath, the floor.

 

Fig 2:Simple WLAN

Simple WLAN

The first improvement on this is shown above; a wired backbone network, with a radio link for two PCs, probably portables brought in temporarily. This is a very attractive means of using resources as and when needed. The portables remain portable, while printouts and networking facilities are available to them locally.

WLAN augments wired network

Fig 3: WLAN augments wired network

WLAN augments wired network

More complex systems as shown above use a combination of hard wired and radio techniques. True wandering of the portable units requires provision for hand-off protocols similar to those in mobile phone networks. This serves to complicate the system. In addition, much greater error protection is needed. A typical bit error rate for cables is less than 1 error in 109. Radio links may be 102 or worse. This requires heavy data protection. Use of cyclic redundancy codes may well be insufficient, and more advanced encoding may be necessary, at a cost in useful bandwidth as compared with bandwidth required.


[back to top]

7.3 Frequency Bands Available

  Main bands available are:-

Others may be available locally; different countries have differing regulations, and this area is in constant change. Lower frequencies are less desirable, since antennas tend to become impractical in size for portable equipment. Higher frequencies have some greater path loss, but the lower levels of interfering signals offset this to some extent. Low noise amplifiers are more difficult to manufacture at higher frequencies, but this is relatively unimportant, since RF link budgets for power and signal strength are rarely limited by receiver noise.


[back to top]


7.4 The R.F. Environment

Fig 4 : Multipath

Multipath


It is important to understand the environment in which the RF link is to be located. Ranges will usually be of 'office' dimensions, say 30 metres or less. However, allowance should be made for non-line-of-sight paths, e.g. behind metal filing cabinets. Operation over such paths differentiates radio from optical systems; but the system must actually work in such circumstances. This usually means a transmit power of over 1 W, with a fairly low noise receiver. Although regulatory limits on RF field exposure would permit this in most instances, recent pressure on mobile phone manufacturers to reduce radiation may bring about systems which transmit very much lower powers. In most such systems, there are multiple reflections from within the environment. This is often the saving grace of such systems for the non-line-of-sight paths. Reflections will make the system work. In some cases however, the reflections may be destructive. Imagine a point receiver. If there are several signals arriving from the transmitter, all from different paths and with varying relative phases, then at a single frequency it may be that the net signal is zero. This must be avoided in a general-purpose system. In practice, legislation in effect prevents such an occurrence by mandating frequency hopping or direct sequence spread spectrum. A further issue in practical systems is interference. This may take several forms, either the near random noise effect of general interference from say electrical machinery, or the more systematic interference from similar systems on the same frequency band.

Fig 5 :BER and Carrier to Noise ratio

BER and Carrier to Noise ratio

The diagram above illustrates the requirement noted above for a good bit error rate, and relates BER to carrier to noise ratio in a radio channel. Even very good carrier to noise ratio channels are far worse in this respect than cables, although infra-red optical systems can be close to cable noise levels. As a counter-example to this last comment, it is true until an 'interfering' signal occurs, e.g. office furniture or possibly a person moves into the line of sight path. The signal drops to zero, and the bit error rate approaches unity. The path is useless. Similarly for radio links, the maintenance of good paths is essential. The counter example here is the proliferation of radio based security systems e.g. on cars. Actually quite a poor BER is acceptable, since the user is familiar with the process of re-transmitting the data if a fault occurs, i.e. the lock does not open. The data stream required is very short, typically 10s of bits, and re-transmitting is not seen as a problem. Large signals on or near the channel will block the transmission; again there is some user awareness of this limitation. Clearly data between computers is in an entirely different category, since the user expects 'zero' errors over very large data blocks. In a cable, cyclic redundancy codes can deliver this; in a radio system, more complexity is required.


[back to top]


7.5 Combatting Multipath and Interference

  There are several techniques to avoid problems with multi-path signals and interference.
  1. Frequency shift as in frequency hopping. This needs to be "smart" to avoid bad channels from multipath or interferers. However, it is an attractive technique that is similar in many ways to a conventional radio, and is therefore relatively easily understood.
  2. Direct Sequence Spread Spectrum. This spreads the signal over a wide band. In the worst case, multipath only nulls some parts of band, and interfering signals may be eliminated by digital signal processing (DSP) of the signal. It is not true to say there is no interference, but the interfering energy is handled in a different way to that of a conventional radio, and thus may be less of a problem. Other DSSS signal in the same frequency band can co-exist, but ultimately contribute to mutual interference. Past misconceptions about DSSS in respect of co-channel re-use gave rise to many difficulties; it should be understood that DSSS is an alternative to the conventional radio which has some advantages, especially in its realisation in monolithic form. It is not, however, the answer to unlimited digital bandwidth within limited radio bandwidth.


[back to top]

7.5.1 Frequency Hopping

Fig 6 : Frequency Hopping

Frequency Hopping


A basic plan for the frequency hopping is shown above. The sequence of frequencies is usually pseudo-random, although in a band of say 100MHz with 2MHz frequency segments clearly the pattern need not be over-long. System intelligence is needed to skip blocked channels, e.g. due to fixed interferes and due to similar systems. Intuitively, it may be seen that two co-located systems would at best only achieve half the data rate of a single system. Intuitively also, this is unlikely to be achieved in practice.

[back to top]

7.5.2 Packetisation

One of the techniques used to reduce errors in radio systems is 'packetisation':-


The technique is effective, at the cost of data rate.

[back to top]


7.6 Frequency Hopping and Interference

  Probability of Packet Error :-
  Pe = Nim.(J/Nch)

Where J = Number of Frequency Hopping nodes

Nch = Number of available channels

Nim = Number of interference modes For Example, if J=3, and Nch =75, and Nim =1

Pe =1 x (3/75) = 0.04
 

i.e. 4% of packets lost due to collisions.


[back to top]


7.7 Direct Sequence Spread Spectrum DSSS

In the DSSS technique, a Pseudo Random sequence generator "chips" the signal at a high rate. Thus, the spectrum is spread by this rate. Instead of a single frequency 'line' in spectral terms, the energy of the signal is spread over a range of frequencies such that the energy at a given frequency in the range is low. This was originally attractive in military terms, as it has a low probability of intercept. In civil applications, it means that many channels can coexist, although there are limits to this as noted above.

The receiver multiplies the incoming signal by the pseudo-random sequence chip function. This action correlates out wanted signal as shown below.


Fig 7: Bit Inversion Modulation - "Chipping"

Bit Inversion Modulation - Chipping

The diagram above is the key to understanding of DSSS. Note that the diagram has been carefully drawn to represent timing accurately, although relative pulse train lengths have been adjusted for clarity. While it is possible in principle to use an analogue transmission signal, digital signals are much more likely in a real system. This is represented by the top trace. The pseudo-noise (PN) digital sequence is shown in the second trace. These two inputs are 'multiplied' together, using an XOR function, to form the transmitted signal, as in the third trace. The transmission system may be broadly similar to many others, but with the means to spread the signal in response to the transmission data above. This may be at a rate of some hundreds of kHz or more, so the modulation sidebands will spread very widely. The energy in a given bandwidth of the spectrum is therefore very low. The receiver must be broadband, in order to collect the entire transmitted signal. This includes the intermediate frequencies in the receiver; bandwidths are typically hundreds of kHz to tens of MHz. The received signal is then multiplied again by the same PN sequence. A check of the waveforms above will show that the same XOR function produces the original waveform. Clearly both the transmitter and the receiver need much digital hardware, and they are really only practical in monolithic form. In addition, the clock signal for the synchronization must be transmitted in some form. This may be as part of the transmitted signal, or may be by reference to a standard. This then requires re-synchronization circuits to lock the chipping clocks of the receiver and transmitter together.

[back to top]


7.8 Receiver Protection Ratio

Receiver Protection Ratio is a concept that differs from a conventional radio specification, but can ultimately be compared with it. It is derived from the chip rate and the useable bandwidth. In a conventional radio, the selectivity against unwanted signals is produced by the narrowest filters in the system, typically the IF filters or possibly the audio frequency filters ahead of any analogue-to- digital conversion. In a DSSS system, some of this is still valid, but the wide bandwidth precludes high selectivity in this respect. Instead, the concept of processing gain is introduced.


[back to top]

7.8.1 Processing Gain in DSSS

Processing gain takes the place of some or all of the I.F. selectivity in a conventional receiver. Processing gain G is given by:-
 
  G = 10log10(B/R)where B = bandwidth after spreading due to chipping
and R = data rateFor example, if B = 20 MHz and R = 20kHz, G = 30dB

This shows that a further 30dB of protection of the signal against interference is available with the figures shown above. However, if the bandwidth is limited, then so is the processing gain and therefore the possible data rate. Many systems users want a bandwidth much greater than 20kHz; 10MHz, approximately equivalent to 10MBps in a system, is a desirable target. Then if the available bandwidth is 100MHz, the processing gain is only 10dB. The system implication of this is that operation should ideally be free of almost all interference to work well.

Approaching this is a separate but of course related way, we can define the 'Jamming Margin'. This is the level of acceptable interference before the system becomes unusable.

[back to top]


7.9 Jamming Margin

  In a DSSS system, spreading gain comes from the chip rate as indicated above. The jamming margin M is given by:- M = G -(L+C/N)where G = processing gain and L = receiver losses


We define C/N = required carrier to noise for acceptable BER, based on knowledge of system requirements and the graph above.

Suppose that C/N = 11dB, i.e. a BER of 0.01, and G = 20dB and L = 4dB.

Then M = 5dB, which would be acceptable only if no interference was present at all.

In a second example, suppose that C/N = 11dB, G =30dB and L = 4dB

Then M = 15dB, which would be acceptable if there was no strong interference.

The wide band example is worse; for G=10dB and L=4dB, the C/N cannot be better than 6dB, or a BER of only 0.1. Any attempts to beat this by encoding will in general serve only to reduce data rates. What is needed is a reduction in data rate or an increase in bandwidth or both. Practical DSSS WLANs have been used, in the first instances with very low data rates of around 10kHz, and acceptable performance. Attempts to exceed 10MBps have not generally been acceptable, although there are 1 MBps systems available. The rapid improvement in wired systems, to 100MBps or more tends to negate some of the wireless advantages, although the concept will definitely be improved on with time.

[back to top]

7.9.1 Conclusions on DSSS

[back to top]

7.9.2 FH or DSSS?

[back to top]

7.9.3 IEEE 802.11

[back to top]


7.10 Applications for Wireless LANs


[back to top]


7.11 Benefits of WLANs

[back to top]


7.12 Summary

[back to top]


7.13 References

[back to top]


7.14 Comment

Marconi - the first with Spread Spectrum Recent attempts to repeat Marconi's Transatlantic tests failed. The transmitter used was a modern narrow band unit, for licence reasons. The receiver, based on Marconi's design, was wideband - a mismatch. Marconi used a broadband transmitter, spread by the arc quench frequency. This matched his broad band Rx very well. His system was in effect close to modern Spread Spectrum approaches. This illustrates the importance of designing a system through all of its aspects.


[back to top]

7.15 SAQ

Compare and contrast infra-red LANs with radio LANs for a modest office environment, say 2 floors, each of 100m2, mainly open plan, some offices.

What further impact would come from a second building of the same dimensions 30 metres away, wired to the same network. What would be the impact of an unrelated network also 30 metres away?

Build some field strength calculations into a comment on the relative merits of the systems.

[back to top]

7.16 Internet keywords

WLAN, Wireless LAN, local area network, infra-red LAN,

[back to top]

Site Search

Powered by Google
Site Map