what to know about DSSS vs OFDM

Question

In all the detail on DSSS and OFDM, I have missed which characterics of DSSS and OFDM make these distinct technologies. Nothing I've read appears to contrast the two.

As I understand it, both DSSS and OFDM are intended to send many bits in parallel. DSSS spreads data across a channel for redundancy, and the books say that bits are sent in parallel. OFDM divides a channel into "subcarriers" and likewise sends bits in parallel. This might suggest the main difference between the two, that DSSS somehow sends data in parallel on a single carrier, whereas OFDM accomplishes this with multiple carriers. Either way, both take up more bandwidth than 1 signal, so I don't get how they're much different.

What should a network engineer know, about how DSSS and OFDM are fundamentally different from each other?

Update: According to the current CWNA book, OFDM is not a spread spectrum technology.

Answer 1

From what I've gathered, the primary differentiation between DSSS and OFDM comes in a congested or constrained RF environment. Because DSSS is transmitting as many bits as possible all at once, it runs the risk of having some of the transmission disrupted in transit should the RF become less than ideal. In contrast, OFDM sacrifices a bit of performance to transmit the data in those multiple carrier "packets" that allow for reassembly or retransmission in the event of a failure.

If I use an extra-large semi truck to haul freight, it works well so long as the highway doesn't have any obstructions (like narrow lanes or short overpasses). If I break my freight up into a greater number of smaller trucks, I can ensure that some of the freight arrives even if conditions (like a traffic jam) prevent the entire shipment from arriving.

Answer 2

Taken from http://wiki.answers.com/Q/Difference_between_ofdm_dsss_fhss:

Orthogonal Frequency Division Multiplexing (OFDM) is a multiple-carrier (MC) modulation technique which creates frequency diversity. A high-speed data stream is converted into multiple low-speed data streams via Serial-to-Parallel (S/P) conversion. Each data stream is modulated by a subcarrier. That way, instead of having a frequency-selective fading wireless channel, where each frequency component of the signal is attenuated and phase-shifted in different amount, we have multiple flat-fading subchannels.

Spread Spectrum (SS) techniques convert a low-speed data stream into a high-speed data stream. That way, the bandwidth of the modulated carrier becomes much larger than the minimum required transmission bandwidth. This is like Frequency Modulation (FM): we trade transmission bandwidth with Signal-to-Noise (S/N) ratio, meaning that we can have error-free communication transmiting lower-power signals.

The fact is, OFDM is used for 802.11n and upwards, so it seems to be the better of the two. I've been a wireless consultant for a couple of years now, and I've never really been asked any questions regarding modulation of the different specifications. Its good to be aware of the differences, but not much more than that.

Answer 3

An 802.11b DSSS/8CCK (11Mbps) symbol, sometimes called HRDSSS/8CCK, is 8 time multiplexed QPSK symbols created by using 6 bits to select 1 of 64 8 chip (QPSK symbol) base codewords and then each codeword can be given 4 different phases using the other 2 bits DQPSK style^[*] that basically produces the main codeword of 256 possible codewords that are a fixed subset of the 512 possible states of 8 QPSK symbols. The 802.11b DSSS/8CCK symbol has a baud rate of 1.375MBd. The chipping sequence is 8 CCK chips, meaning the sample rate is 11MHz (one sample per QPSK symbol, 1 DSSS/8CCK symbol per baud), so the channel is 22MHz wide. The information rate is therefore 5.5Mbps with 4CCK and 11Mbps with 8CCK, with gross bit rate and coded data rate (chipping rate) both 11Mbps when using 4CCK and 22Mbps when using 8CCK (because 8 bits are spread to 16 bits). The 1Mbps (DSSS/Barker1) / 2Mbps (DSSS/Barker2) modes used 11 chip Barker at 1MBd and used 1 DBPSK / DQPSK symbol in the DSSS symbol multiplexed with the barker code, you end up with 11 time multiplexed QPSK symbols in the DSSS/Barker2 symbol. 802.11b also has a non DSSS mode called PBCC-11, which uses a coding rate of 1/2 and is a single QPSK symbol transmitted at 11MBd, and PBCC-5.5 is the same but is a single BPSK symbol. In the above nomenclature we see that 8CCK and PBCC are both combined coding and modulation schemes, QPSK is a modulation scheme and DSSS indicates that 8CCK is a direct spread spectrum technique.

^[*] DQPSK style because DQPSK takes 2 bits from the bitstream, which selects a phase offset from the previous QPSK symbol it transmitted, which it then adds to the phase of the previous symbol to get the phase of the QPSK symbol. CCK takes 2 of 8 bits from the bitstream, which selects a phase for the whole DSSS/8CCK symbol, which is taken as an offset from the phase of the final QPSK symbol in the previous DSSS/8CCK symbol, which happens to be the phase that was chosen using the DQPSK at the start (first 2 bits) of the previous symbol, because the last QPSK symbol in that DSSS symbol is e^φ₁. The formula for generating the 8 QPSK symbols in 8CCK is as follows:

Where φ₁ is determined by the first 2 bits of an 8 bit chunk in the bitstream using DQPSK and φ₁, φ₂ and φ₃ are encoded as absolute phase using the phase table, where the next 2 bits determine φ₂, and so on, and is an absolute phase and not an offset from anything. These then produce different 8 QPSK symbol codewords according to the generator above.

An 802.11a OFDM symbol consists of 80 subcarriers (48 of which are useful data), so 80 QAM16 (QAM16 is used in 36Mbps mode) symbols frequency multiplexed over the subcarriers (one per subcarrier). The baud rate of the OFDM symbol, and therefore every subcarrier within it, is 250KBd, and when each subcarrier is modulated with QAM16, the OFDM symbol has 48*4 = 192 bits of useful data, so you get an information rate of 192 bits*250KBd*0.75 = 36Mbps, if you used a coding rate of 3/4, and a coded data rate therefore of 48Mbps, and a gross bitrate of 80Mbps when you include the CP and pilot subcarriers. The spectral efficiency of a single subcarrier is actually 4bpcu when it contains useful data, but the average spectral efficiency of the QAM16 symbols as a whole is 1.8bpcu, meaning the OFDM symbol has a spectral efficiency of 144bpcu. In bps/Hz this is 1.8bps/Hz due to the 250KBd per 20MHz i.e. 0.0125 Bd/Hz so (0.0125*144) bps/Hz. The spectral efficiency of a DSSS/8CCK symbol is 8bpcu, but in bps/Hz, is 0.5bps/Hz due to 1.375MBd per 22MHz i.e. 0.0625Bd/Hz.

I therefore conclude that 802.11a OFDM/QAM16 3/4 coding rate is 3.6x more spectrally efficient than 802.11b DSSS/8CCK (where chip i.e. coding rate is inherently 1/2). It manages a 3.27x greater information rate with a 1.1x smaller bandwidth.