Overview on synchronization in OFDM systems

Orthogonal Frequency Division Multiplexing (OFDM) is a advance bandwidth efficient digital modulation technique highly adopted for the broadband wireless services because it gives high performance in the multipath fading environments. It is very robust against narrowband interference, but it is not totally free from drawbacks. There are timing and frequency synchronization error for which this technique is very sensitive. These errors are quite detrimental to the mutual orthogonality among the subcarriers which is main condition of the OFDM technique. If it is not resolved properly, it creates inter-symbol interference (ISI) and inter-channel interference (ICI). ISI and ICI increase the Bit Error Rate (BER) and creates packet loss in the transmission. Means, OFDM System has no proper use without precise synchronization technique to assure reasonable system performance. This paper provides an overview on the causes and the effects of these synchronization errors. It focuses on various symbol timing and carrier frequency acquisition schemes including coarse and fine synchronization techniques, using cyclic prefix or training symbols. It has also discussed the attention to implement different standards in the same wireless device using common frequency synchronization structure.


INTRODUCTION
Future era of wireless communication would be of wireless systems such as high speed internet, multimedia services and streaming video (Digital Video Broadcasting) are needed to meet the high speed, good quality and better mobility requirements and also reasonable spectrum efficiency.It is impossible to implement such systems using conventional serial communication in frequency-selective fading channel.Parallel communication concept means multicarrier modulation scheme is must to attain high speed services in multipath fading environments.Presently, most of the broadband wireless communication systems are based on OFDM as it turns out to be a strong technique due to its high spectral efficiency and simplicity in equalization [26].In OFDM, available broad channel bandwidth is split into adjacent narrow band channels, and the high rate data stream is split into several low data rate streams which are multiplexed to the orthogonal subcarriers and transmitted simultaneously.
OFDM has been widely adopted in broadband wireless communication systems that include wireless local area networks (WLAN/IEEE 802.11a/g/n/ac and HIPERLAN/2) [1], [2], [3], [4], wireless metropolitan area networks (WMAN/WiMax, IEEE 802.16) [5], terrestrial digital audio broadcasting (DAB) and terrestrial digital video broadcasting (DVB) systems [6], [7] due to its robustness against the effects of multipath fading.The third generation (3G) systems which are developed to attain high data rate services at 2Mb/s for fixed users and 384kb/s for mobile users in mobile internet communication.The fourth generation (4G-LTE) systems [8] will support even more demanding physical layer requirements, such as streaming video.
Though OFDM offers higher performance and benefits over the traditional single carrier modulation techniques and achieves spectrally efficient high data rates close to the Shannon limit, it is not totally free from hurdles.The main challenges in the use OFDM technique are high peak-to-average power ratio (PAPR) and sensitivity to synchronization errors.High value of PAPR creates hurdle in the implementation of analog to digital converter (ADC) and digital to analog converter (DAC).It also needs the linear power amplifier which has less power efficiency because PAPR of OFDM is proportional to the number of subcarriers used in OFDM systems.Due to multicarrier transmission, time and frequency synchronization between the transmitter and receiver are very important to maintain in limit for the accurate detection of the signal at the receiver and set up quality link [10], [11].Error in both timing and frequency synchronizations introduce additional interference to the OFDM systems which ultimately degrade the performance [26].Proper channel estimation is key part of any communication link which is severely affected by timing synchronization in OFDM [9].A wide variety of techniques have been proposed for the effective estimation and correction of both timing and carrier frequency errors at the OFDM receiver.It is necessary to reduce computation complexity of synchronization part while maintaining reasonable synchronization performance as synchronization part in the OFDM system is the second next to Fast Fourier Transform (FFT) calculation in computational complexity.
The main objective of this paper is to discuss the synchronization issues for OFDM systems and focus some insights into the causes, effects and means of reducing them.Section II covers the basics of synchronization.Section III and Section IV discuss time synchronization and carrier frequency synchronization techniques respectively.Paper ends with some conclusions in the last section V.

II. SYNCRONIZATION
Synchronization system is the heart of the overall OFDM system.Accuracy of the frequency and timing error estimation and correction decides the performance of OFDM system.The effects of carrier frequency and timing errors are different over different channels and it is very important to study these effects before to set up OFDM link [12].
Timing synchronization issue can be divided in two parts: symbol synchronization and sampling clock synchronization.Multipath effect of the wireless channel spreads the OFDM symbol which creates error in the detection of the starting of the symbol.This is called inter-symbol-interference (ISI).ISI also occurs due to sampling clock offset because it generates drifting of the OFDM symbol block away from the receiver FFT window.These two phenomena lead to ISI which then creates inter-carrier-interference (ICI).ISI and ICI together disturb the accuracy of detection system at the receiver and thus degrade the performance of the system.Timing synchronization also affects the channel estimation in OFDM link.Channel estimation is carried out by evaluating channel impulse response (CIR) which is to be within the time domain estimation window but timing offset shifts the location of the CIR.Shifting of CIR disturbs the channel estimation.
Deviation in the carrier frequency of local oscillator at transmitter and receiver plays key role in the performance of the OFDM system.This deviation shifts the frequency spectrum of the received signal which in result of ICI due to loss of orthogonality among subcarriers.Subcarriers remain orthogonal if frequency error is an integer multiple k of the subcarrier spacing f Δ .In this case, the spectrum of the received signal is shifted by f kΔ which also increases bit error rate (BER) because of occurrence of received data symbols at the wrong position in the demodulated spectrum.In short, frequency error creates inter-carrier-interference in either case and finally BER increases.Without solving the issues of timing and frequency error, Error free OFDM system is not possible.Many multimedia application adopted OFDM modulation technique so the synchronization algorithms used to undertake the problem of timing and frequency error depend upon the application.It is also under research to finalize common algorithm to be used for handling the issue of synchronization.Common synchronization system is essential because of today's demand to employ all standards having OFDM concept in the same wireless device.[13].Mostly and many authors had suggested that coarse timing and frequency acquisition done by known pilot symbols called preamble or pilot tones embedded into the OFDM symbols.[14], [15], [16].On other hand, cyclic extension concept was also proposed for undertaking the issues of fine frequency and timing synchronization [16], [17], [18].In short, proper and effective synchronization algorithms have been investigated in order to improve the performance of the OFDM system against timing and frequency errors.
Looking to the different standards finalized for the OFDM system, it is found that different methods are being used for handling the issue of synchronization.In 802.11a standard, synchronization problem is resolved by using preamble part embedded in the main frame of the OFDM System.802.11n and 802.11ac use high throughput (HT) preamble along with legacy preamble to provide backward compatibility for the 802.11a standard.This is regular leaning in the development of particular standard so that devices of new standard can be used in the old technology also.Added HT preamble in the 802.11n standards look for the channel estimation and fine-tuning of AGC control in Multi-Input-Multi-Output OFDM system.Ten identical short training symbols and two identical long training symbols plus guard interval are used as legacy preamble for reducing the synchronization errors.Each additional symbol act upon specific function such as short training symbols are used for coarse timing synchronization while fine timing synchronization can be done by long training symbols.Auto-correlation and crosscorrelation based algorithms are popular for cancelling timing synchronization errors.It is required to note down that conventional auto-correlation based algorithms have a plateau problem and conventional cross-correlation based algorithms have pseudo multipath problem caused by Cyclic Delay Diversity (CDD) [30].
Long Term Evolution (LTE) -4G employs pilot subcarriers instead of preamble symbols to make easy the process to cancel the timing and frequency synchronization errors.Similarly, DVB -T/H also utilize dedicated subcarriers for timing and frequency synchronization purposes. .

III. TIMING SYNCRONIZATION
Erroneous timing synchronization causes inter-symbolinterference and inter-carrier-interference which badly deteriorates BER of the system [9].Generally, timing synchronization is performed in two parts: Rough timing synchronization and fine timing synchronization.Timing synchronization is applied in OFDM system depends on type of application because structure of the frame is not same in all applications.Short distance communications like Wireless Local Area Networks make use of preambles while Digital Video/Audio broadcasting system use many symbols to facilitate the solution against synchronization error.
While handling the synchronization issue, it is suitable and more practical to tackle synchronization error in time domain rather in frequency domain.In practice, initially carrier frequency offset is not performed so that cross correlation based algorithms cannot be used.As mentioned in the above paragraph, the rough/coarse timing synchronization is done first.There are two ways to handle the process of coarse timing synchronization which depend on the requirements of the OFDM systems.Based on application, coarse timing synchronization error is looked after with the help of cyclic prefix/suffix [19], [20], [21], [22] or by the dedicated preamble [23], [24].Both algorithms have their own characteristics as cyclic prefix based methods can work blindly whereas algorithms based on preamble can generally gives faster and more robust output against error of synchronization.It is reviewed that the maximum likelihood (ML) [21], minimum mean squared error (MMSE) [20] and the maximum correlation (MC) methods [19], [25] are three basic methods to resolve timing synchronization.
Maximum correlation method as suggested by Hao Zhou et al. [26] is employed in the case of training sequence in which the correlation metric is maximized .
Where N is the length of the repeated pattern in the preamble and p is the received time domain sequence with a time index as subscript.
The MMSE method can be expressed as maximizing is the same as in equation 1 and The ML method is to maximize With the same ) ( j Z and ) ( j R as defined above and There is one more method proposed by Schmidl and Cox [23] in addition to above mentioned methods.They suggested an estimator that maximizes )) ( ( With OFDM modulation technique is very useful in multipath channel for high speed communication.Statistical information of multipath channel can be employed in ML coarse symbol synchronization algorithm [22] and ML timing scheme for fast Rayleigh fading channel is very nicely explained by Lin [27].It is also way to optimize the preamble using method of false detection under low SNR that is based on MMSE normalized metric method proposed by Minn [24]. Training sequence based algorithms are very easy and also having low complexity of implementation.But long repetition structure creates well known plateau problem.The effect of this problem is resulted into large estimation error [28], [29].Plateau problem can be avoided in the cross correlation based algorithms at the cost of computational complexity.Cross correlation based algorithms give better result in Additive White Gaussian Noise (AWGN) channel than multipath channel.Pseudo multipath problem may take place due to multipath delay.Multipath delay is one of the sources to diestrum synchronization among subcarriers in OFDM link.
Many authors proposed effective algorithms over and above mentioned auto-correlation and cross correlation based algorithm.Three step timing synchronization scheme having sliding window differentiator, SIR metric formation and frame timing refinement process [30] solved the plateau problem of auto-correlation based algorithm and pseudo multipath problem of cross-correlation based algorithms.Slinging window process decreases the variance of estimation in coarse timing of frame which then removes the plateau problem.After coarse timing synchronization without plateau problem, exact timing synchronization is achieved by defining a metric called signalto-interference ratio (SIR).Gaussian approximation of interference which is resulted from timing error is taken as input while defining SIR metric.At the last, frame timing is refined to eliminate timing synchronization error.This is very effective method for timing synchronization.
Three step timing synchronization is initiated by detecting the peak of is having plateau due to long training sequence of legacy preamble.It creates confusion because position of peak on the plateau is uncertain.To get perfect position of peak a sliding differentiator is used after the auto-correlator, which gives output as Starting time estimation can be done by noticing the exclusive peak of . From this estimation, coarse frame and symbol timing can be derived.
With the help of signal-to-interference-ratio (SIR) metric calculator, the fine timing synchronization is obtained.SIR metric calculator is concatenated to the cross correlation synchronizer in order to calculate Where Narae Kim proposed antenna selection algorithm [31] which decreases the power requirement of the previous three step method.It estimates the received powers of different antenna and selects the signal having high power for the synchronization process.Signal detection is done by delay and correlation algorithm [32].Adaptive accumulation window concept with antenna selective timing synchronization for MIMO-OFDMbased IEEE 802.11ac wireless LAN systems also works [40].
Sampling frequency can be adjusted in reference to fine timing procedure to reduce the sampling clock offset.[19], [24], [34].There is a parallel processing scheme in which sampling clock offset and carrier frequency offset estimated simultaneously by phase rotation of successive OFDM symbols [9], [33].
IV. FREQUENCT SYNCRONIZATION There are many reasons such as temperature changes, oscillator frequency drift (Doppler spread effect), tolerance of RF components, etc. responsible for the mismatch in carrier frequencies of the transmitter and the receiver.Carrier frequency offset (CFO) introduces shift in the spectrum of the received OFDM symbols which finally creates inter-carrier-interference due to loss of orthogonality among subcarriers.If carrier frequency offset is not nullified, it creates great number of interfering subcarriers at the receiver which finally decreases bit error rate.
It is reviewed that the number of subcarriers does not influence the ICI noise variance for OFDM symbol lengths of more than 64 subcarriers.It is fact the interference level is quickly down as frequency partition among subcarriers increasing.It means closed subcarriers add significant load of interference on the subcarriers [12].Carrier frequency offset also down the received signal amplitude notably along with ICI.It is also found that the effect of CFO depends on type of modulation techniques used OFDM system.Coherent detection is less influenced by CFO than non coherent detection.Modulation based CFO discussion and the effect of CFO has been very well explained [10], [35], [36].CFO also deteriorates bit error rate in AWGN channel [10] and in Rayleigh fading channels [35] which can be approximated.Reduction in signal-to-noise ratio due to ICI is evaluated by deriving the ICI power [36].
Like timing synchronization, CFO is also compensated in steps means first coarse frequency synchronization and then fine frequency synchronization.Generally, coarse frequency synchronization means frequency acquisition algorithms provide sufficient accuracy.Initial estimate of frequency error must be precise to half subcarrier spacing as it is essential for reliable fine tuning process.Frequency shift of received signal can be easily made known by detecting the drift of location of pilot tone.Algorithm based on pilot tone in OFDM data symbols, surrounded by zero valued virtual subcarriers can make easy to locate shifted pilot at receiver [15].It is also possible to tackle CFO by using repeated data symbols [16] and by adding cyclic extension in OFDM symbols [17], [18].In this method, the phase shift between received data samples and their repeated copies used in cyclic extension is evaluated.Carefully designed training symbols support in availing coarse and fine carrier frequency along with timing synchronization also [24].
All above methods of CFO synchronization, ML estimator is used which is very complex.It is very complex to estimate the frequency error in fast time varying channels using ML estimator.Excellent approach has been accepted to estimate the CFO by analyzing statistically the fading process due to fast time varying channels.[37].This method has used time domain channel estimation concept to minimize complexity of ML estimator.There are few algorithms performed in frequency domain by deducing the weights for weighted least square sampling clock offset and CFO estimator.Estimating [33] and same way the residual carrier and sampling frequency offset are trailed using the temporal correlation in the data aided post fast fourier transform stage.In [38], authors proposed algorithm can work in the presence of narrow band interference which considerably complicates the synchronization process.It has been nicely explained the process of CFO recovery in OFDM based cognitive radio systems overwhelmed by narrow band interference.
Obviously, different methods for CFO recovery are used in different applications so it is essential to compare the different synchronization methods for designing common wireless device.Four standards 802.11n, 802.16d,LTE, and DVB-T/H are taken into discussion of carrier frequency synchronization and also explain relationship between different methods [13].LTE and DVB-T/H and cellular systems are circuit switched while 802.16d and 802.11n are packet switched systems, so their frequency synchronization algorithms are different as per requirements.It has discussed synchronization methods application wise and concluded best solution for the common device structure.
Studying different methods of carrier frequency synchronization, it is clear that all the algorithms fulfill the estimation accuracy of half subcarrier spacing which is necessary if subsequent fine frequency tracking mechanisms are used data symbol by data symbol.But if coarse frequency synchronization done by strict constraints on the estimation of frequency offset, there is no need of fine tuning.[39].
V. CONCLUSION We have discussed very crucial part of the OFDM system that is timing and frequency synchronization.Due to timing and frequency offset, ISI and ICI occurs which reduce the signal power, SNR degradation and finally deteriorates the performance of the system.OFDM has no use if timing and frequency errors are not estimated properly and compensated subsequently.We surveyed various existing methods for coarse and fine synchronization to reduce the effects of the timing and frequency offset.Cyclic prefix based synchronization schemes can be applied without preamble and preamble based schemes can achieve faster and more robust performance on OFDM systems.Different algorithms for different standards of application are discussed with reasons.For example, in the application like WLAN, coarse synchronization may be sufficient while in DVB, fine synchronization is necessary to correct the residual errors after coarse synchronization.Multistandard synchronization technique with less complexity and resources is also briefly discussed.

2 2
of Guard Interval, the length of the window of power signal which satisfies g e X X ≤ and the length of the obstruction power windows, respectively.Three step algorithm detect the peak of ) (i d Ω instead of peak of d λ to determine the start point of guard interval of long training sequences in legacy preamble.In a very minute way, finally, refine timing synchronization is by very small search window around the estimation of the previous step.This new algorithm gives excellent timing synchronization compared to conventional auto correlation and cross correlation schemes.But it uses more power due to complexity and threshold dependent performance while availing synchronization.