BLAST (Bell labs Layered Space Time Technology) Multipath Technology

ABSTRACT

In wire less transmission the radio waves do not simply propagate from transmit to receive antenna, but bounce and scatter randomly off objects in the environment. This scattering is known as multipath as it results in multiple copies of transmitted signals, arriving at the receiver via different scattered paths. In conventional wireless systems, multipath represents a significant impediment to accurate transmission, because the images arrive at the receiver at slightly different times and can thus interfere destructively, canceling each other out. For this reason, multipath is traditionally viewed as a serious impairment. A layered space time technology by Bell labs to exploit the concept of multipath known as BLAST (Bell labs Layered Space Time Technology). Using the BLAST approach however, it is possible to exploit multipath, that is, to use the scattering characteristics of the propagation environment to enhance, rather than degrade, transmission accuracy by treating the multiplicity of scattering paths as separate parallel sub channels. By this method the spectrum is used more efficiently.

INTRODUCTION

The explosive growth of both the wireless industry and the internet is creating a huge market opportunity for wireless data access .Limited internet access, at very slow speeds, is already available as an enhancement to some existing cellular systems. However those systems were designed with purpose of providing voice services and at most short messaging, but not fast data transfer. Traditional wireless technologies are not very well suited to meet the demanding requirements of providing very high data rates with the ubiquity, mobility and portability characteristics of cellular systems. Increased use of antenna arrays appears to be the only means of enabling the type of data rates and capacities needed for wireless internet and multimedia services. While the simultaneous deployment of base stations and terminal arrays that can unleash unprecedented levels of performance by opening up multiple spatial signaling dimensions. Theoretically, user data rates  as high as 2Mb/sec will be supported in certain environments, although recent studies have shown that approaching those might be feasible under extremely favorable conditions-  in the vicinity of the base station and with no other user s competing for bandwidth .Some  fundamental barriers related to nature of radio channel as well as to limited band width availability at the frequencies of interest stand in the way of high data rates and low cost associated with wide access.

FUNDAMENTAL LIMITATIONS IN WIRELESS DATA ACCESS

Ever since the dawn of information age, capacity has been the principal metric used to assess the value of a communication system. Since the existing cellular systems were devised almost exclusively for telephony, user data rates were low. In fact the user data were reduced to a minimum level and traded for additional users. The value of a system is no longer defined only by how many users it can support, but also by its ability to provide high peak rates to individual users. Thus in the age of wireless data, user data rates surges as an important metric.

Trying to increase the data rates by simply transmitting more; Power is extremely costly.  Furthermore it is futile in the context of wherein an increase in everybody’s transmit power scales up both the desired signals as well as their mutual interference yielding no net benefit. 

Increasing signal bandwidth along with the power is a more effective way of augmenting the date rate.  However ratio spectrum is a scarce and very expensive resource. Moreover increasing the signal bandwidth beyond the coherent bandwidth of the wireless channel results in frequency selectively.  Although well-established technique such as equalization and OFDM can address this issue, their complexity grows with the signal bandwidth. Spectral efficiency defined as the capacity per unit bandwidth has become another key metric by which wireless systems are measured. The entire concept of frequency reuse on which cellular systems are based constitutes a simple way to exploit the spatial dimension. Cell sectorisation a wide spread procedure that reduces interference can also be regarded as a form of spatial processing.

LIFTING THE LIMITS WITH TRANSMIT AND RECEIVE ARRAYS

In wireless systems, radio waves do not propagate simply from transmit antenna to receive antenna, but bounce and scatter randomly off objects in the environment. This scattering is known as multipath, as it results in multiple copies ("images") of the transmitted signal arriving at the receiver via different scattered paths. In conventional wireless systems, multipath represents a significant impediment to accurate transmission, because the images arrive at the receiver at slightly different times and can thus interfere destructively, canceling each other out. For this reason, multipath is traditionally viewed as a serious impairment. Using the BLAST approach however, it is possible to exploit multipath, that is, to use the scattering characteristics of the propagation environment to enhance, rather than degrade, transmission accuracy by treating the multiplicity of scattering paths as separate parallel sub channels.

OVERVIEW OF BLAST SYSTEM:

The prevailing view was that each wireless transmission needed to occupy separate frequency, similar to the way in FM radio with in a geographical area allocated with separate frequencies. BLAST technology essentially exploits a concept that other researchers believed impossible. The original scheme developed was D BLAST, which utilizes multi-element antenna arrays at both transmitter and receiver and an elegant diagonally layered coding structure in which code blocks are dispersed across diagonals in space-time. In an independent Rayleigh scattering environment, this processing structure leads to theoretical rates which grow linearly with the number of antennas (assuming equal numbers of transmit and receive antennas) with these rates approaching 90% of Shannon capacity. Scattering of light off the molecules of the air, and can be extended to scattering from particles up to about a tenth of the wavelength of light. Rayleigh can be considered to be elastic scattering because the energies of scattered photons do not change. The coding sequence used in D BLAST is very complex and costly. So we move to the most current iteration V BLAST. 

LST code encoding process. Here, n= 3. (a) The Incoming information bit sequence is first demultiplexed into n subsequences. Each subsequence is then encoded using a constituent code. (b) The coded symbols from the CCs are

transmitted by the n transmitting antennas in turn. 

MOST CURRENT ITERATION-V BLAST

A single data stream is de multiplexed into M sub streams .Each sub stream is then encoded into symbols and fed to its respective transmitter. Transmitters 1 through operate co channel at symbol rate 1/ T symbols/sec, with synchronized symbol timing. Each transmitter is itself an ordinary QAM transmitter. QAM combines phase modulation with AM. Since the entire sub streams are transmitted in the same frequency band, spectrum is used very efficiently. Since the user’s data is being sent in parallel multiple antennas are used. QAM is an efficient method for transmitting data over limited bandwidth channel. It is assumed that the same constellation is used for each sub stream, and that transmissions are organized into bursts of L symbols. The power launched by each transmitter is proportional to 1/ M so that the total radiated power is constant irrespective of the number of transmitting antennas. Blast’s receivers operate co channel, each receiving signals emanating from all M of the transmitting antennas .It is assumed that channel-time variation is negligible over the symbol periods in a burst.

BLAST’S SIGNAL DETECTION

At the receiver, an array of antennas is again used to pick up the multiple transmitted sub streams and their scattered images. Each receiving antenna "sees" the entire transmitted sub streams superimposed, not separately. However, if the multipath scattering is sufficient, then the multiple sub streams are all scattered differently, since they originate from different transmit antennas that are located at slightly different points in space. Using sophisticated signal processing, these differences in scattering of the sub streams allow the sub streams to be identified and recovered. In effect, the unavoidable multipath in wireless communication offers a very useful spatial parallelism that is used to greatly improve bit-rates. Thus, when using the BLAST technique, the more multipath, the better, just the opposite of conventional systems.

The BLAST signal processing algorithms used at the receiver are the heart of the technique. At the bank of receiving antennas, high-speed signal processors look at all the signals from all the receiver antennas simultaneously, first extracting the strongest sub stream from the morass, then proceeding with the remaining weaker signals, which are easier to recover once the stronger signals have been removed as a source of interference. Again, the ability to separate the sub streams depends on the differences in the way the different sub streams propagate through the environment.

Let us assume a signal vector symbol with symbol-synchronous receiver sampling and ideal timing. If a = ( a1,a2,a3,….am )T is the vector transmitted symbols, then receiver N vector is R1=Ha+V, where H is the matrix channel transfer function and V is a noise vector.

Signal detection can be done using adaptive antenna array techniques, some times called linear combinational nulling. Each sub stream is sequentially understood as the desired signal. This implies that the other sub stream will be understood as interference. One nulls this interference by weighting signals they go to zero (known as zero forcing).

While these linear nullings works, on linear approaches can be used in conjunction with them for overall result. Symbol cancellation is one such technique. Using interference from already detected components of interfering signals are subtracted to form the received signal vector. The end result is a modified receiver vector with little interference present in the matrix. Bell labs actually tried both approaches. The result showed that adding the non linear to the linear yielded the best performance and dealing with the strongest channel, first (thus removing it as interference) give the best overall SNR. If all components of ‘a’ are assumed to be the part of the same constellation, it would be expected that the component with the smallest SNR would dominate the overall error performance. The strongest channel then becomes the place to start symbol cancellation. This technique has been called the “best first” approach and become the de-facto way to do signal detection from an RF stream. But what the Bell labs guys found is that if you evaluate the SNR function at each stage of the detection process, rather than just at the beginning, you come up with a different ordering that is also (minimax) optimal.

As its core V BLAST is an iterative cancellation method that depends on computing a matrix inverse to solve the zero forcing function. The algorithm works by detecting the strongest data stream from the received signal and repeating the process for the remaining data streams. While the algorithm complexity is linear with the number of transmitting antennas, it suffers performance degradation through the cancellation process. If cancellation is not perfect, it can inject more noise into the system and degrade detection.

The essential difference between D BLAST and V BLAST lies in the vector encoding process. In D BLAST, redundancy between the sub streams is introduced through the use of specialized inter-sub stream block coding. In this code blocks are organized along diagonals in space time. It is this coding that leads to D BLASTS higher spectral efficiencies for a given number of transmitters and receivers. In V BLAST, however the vector encoding process is simply a demux operation followed by independent bit-to-symbol mapping of each sub stream. No inter-sub stream coding, or coding of any kind is required, though conventional coding of the individual sub streams may certainly be applied.

LABORATORY RESULTS

A laboratory prototype of a V BLAST system has been constructed for the purpose of demonstrating the feasibility of the BLAST approach. The prototype operates at a carrier frequency of 1.9 GHZ and a symbol/sec, in a band width of 30 KHz. The system was operated and characterized in the actual laboratory office environment not a test range, with transmitter and receiver separations up to about 12 meters. This environment fading is relatively benign in that the delay spread is negligible, the fading rates are low and there is significant near-field scattering from near by equipment and office furniture. Nevertheless, it is a representative indoor lab/office situation, and no attempt was to “tune” the system to the system to the environment, or to modify the environment in anyway.

The antenna arrays consisted of λ/2 wire dipoles mounted in various arrangements. For the results shown below, the receive dipoles were mounted on the surface of a metallic hemisphere approximately 20 cm in diameter, and transmit dipoles were mounted on a flat sheet in a roughly rectangular array with about λ/2 inter-element spacing. In general, the system performance was found to be nearly independent of small details of the array geometry.

Fig. 7.3 shows the results obtained with the prototype system, using M=8 transmitters and N=12 receivers. In this experiment, the transmit and receive arrays were each placed at a single representative position within the environment, and the performance characterized. The horizontal axis is spatially averaged receiver SNR. The vertical axis is the block error rate, where a “block” is defined as a single transmission burst. In this case, the burst length L is 100 symbol duration of which is used for training. In this experiment, each of the eight sub streams utilized uncoded 16 QAM, ie. 4 bits/symbol/transmitter, so that the payload block size is 8*4*80=2560 bits. The spectral efficiency of this configuration is 25.9 bps/Hz and the payload efficiency is 80% of the above or 20.7 bps/Hz, corresponding to a payload data rate of 621 Kbps in 30 KHz band width.

The upper curve in fig. shows performance obtained when conventional nulling is used. The lower curve shows performance using nulling and optimally-ordered cancellation. The average difference is about 4 db, which corresponds to a raw spectrally efficiency of around 10 bps/Hz.

Fig. 7.4 shows performance results obtained using the same BLAST system configuration (M=8, N=12, 16-QAM) when the receive array was left fixed and the transmit array was located at different positions throughout the environment. In this case, the transmit power was adjusted so that large received SNR was 24+/-0.5 db. Nulling with optimized cancellation was used.

It can be seen that at this spectral efficiency is reasonably robust with respect to antenna position. In all positions, the system had at least 2 orders of magnitude margin relative to 10^-2 BER. For a completely uncoded system, these are entirely reasonable error rates, and application of ordinary error correcting codes would significantly reduce this. At 34 db SNR, spectral efficiencies as high as 40 bps/Hz have been demonstrated at similar error rates, though with less robust performance.

BLAST IN THE REAL WORLD

Two familiar factors are there for the success of BLAST: technology and economics. On technology side, scalar systems (those currently in use) are far less spectrally efficient than BLAST ones. They can encode B bits per symbols using a single constellation of 2B points. Vector systems can realize the same rate using M constellation of 2B/M points each. That is large spectral efficiencies are more practical. Let’s take an example. If you want 26 bps/Hz with a 23% roll off, you need to have (26*1.23) = 32 bits/symbol. A scalar system would require 232 points, which is around 4 billion. No wireless system will put up 4 billion transmitters ever. This means that the vector approach is the only one that one can ever hope to fulfill such a bit-per-second rate. On the economic side, BLAST calls for an infrastructure that will take considerable resource to develop. Cell antennas will have to be redesigned to evolve with the increase in data rates. The first change will have to occur at the cell towers, and then at the receivers. The cell tower will have to go from a switched-beam approach to a steered beam configuration. On the plus side, much of this development can be gradual. Older "diversity" antennas will most likely be retained as a fallback for the worst-case channel environments (which means single-path flat-fading at low mobile speeds), so new antennas can be added gradually. A carrier could go from one to two to four transmit paths per sector, upping the cost of service with each incremental performance gain. Proceeding with a hardware-based migration will yield balanced gains in the forward and reverse links. Carriers are very sensitive to the costs, however incremental, of deploying new systems.

BLAST VS.  EXISTING SYSTEMS

What makes BLAST different from any other single –user that uses multiple transmitters? After all, we can always drive all the transmitters using a single user’s data, even it is sub streams. Well, unlike code-division or a speed spectrum approach, the total bandwidth those QAM systems require. Unlike a Frequency Division Multiple Access (FDMA) approach, each transmitted signals occupies the signal bandwidth. And finally, unlike Time Division Multiple Access (TDMA), the entire system bandwidth is used simultaneously by all of the transmitters all of the time. Blast system does not impose orthogonalisation of transmitted signals. The reason for this is simple, obvious and rather elegant. The propagation environment of the real world provides significant latencies. BLAST exploits them to provide the signal décor relation necessary to separate the co-channel signals .BLAST uses the same effect that cause ghosting in TV pictures as a sort of clock to allow the various signals to be extracted.

ADVANTAGES

Since the entire sub streams are transmitted in the same frequency band, the spectrum is used efficiently. Spectral efficiency of 30-40 bps/Hz is achieved at SNR of 24 db. This is possible due to use of multiple antennas at the transmitter and receiver at SNR of 24 db. To achieve 40bps/Hz a conventional single antenna system would require a constellation with 10^12 points. Further more a constellation with such density of points would require in excess of 100 db operating at any reasonable error rate.

A critical feature of BLAST is that the total radiated power remains constant irrespective of the number of transmitting antennas. Hence there is no increase in the amount of interference caused ton users.

The BLAST technology has reportedly delivered a data reception at 19.2 Mbps on a 3G network. With BLAST down loading a song would take 3s, and HDTV can be watched on a telephone.

This innovation known as BLAST may allow so called “fixed” wireless technology to rival the capabilities of today’s wired networks would connect homes and business to copper-wired public telephone service providers.

                                     

CONCLUSION

Under widely used theoretical assumption of independent Rayleigh scattering theoretical capacity of the BLAST architecture grows roughly, linearly with the number of antennas even when the total transmitted power is held constant.  In the real world of course scattering will be less favourable than the independent Raleigh’s assumption and it remains to be seen how much capacity is actually available in various propagation environments.  Nevertheless, even in relatively poor scattering environment, BLAST should be able to provide significantly higher capacities than conventional architectures.