PaperlayerView exact revision on arXiv

Source-linked AI summary

Flip-OFDM for Optical Wireless Communications

Nirmal Fernando, Yi Hong, Emanuele Viterbo

arXiv:1111.5682v1cs.ITmath.IT

TL;DR

Optical wireless OFDM faces real, positive-signal constraints, and Flip-OFDM had received little study compared with ACO-OFDM. The paper analyzes Flip-OFDM and modifies it for a fair comparison; simulations report matching spectral efficiency and electrical-domain BER, with lower receiver complexity for Flip-OFDM in slow fading.

  • Problem

    ACO-OFDM had been widely studied, while Flip-OFDM had been largely ignored despite optical wireless OFDM's real, positive-signal constraints.

    Source: §I · p. 1

  • Method

    The paper analyzes Flip-OFDM, modifies it to match ACO-OFDM's bandwidth and cyclic-prefix conditions, and compares performance and complexity.

    Source: §I · p. 1, A. Modification of Flip-OFDM · p. 3

  • Results

    Both schemes have the same spectral efficiency and electrical-domain BER, while Flip-OFDM saves 50% in receiver complexity for slow fading.

    Source: A. BER · p. 4, §V · p. 5, B. Complexity of TX and RX operations · p. 5

  • Takeaways & Limitations

    For slow-fading channels, Flip-OFDM is a lower-receiver-complexity alternative to ACO-OFDM with matching reported spectral efficiency and electrical-domain BER.

    Source: §V · p. 5

Abstract

from arXiv · show

We consider two uniploar OFDM techniques for optical wireless communications: asymmetric clipped optical OFDM (ACO-OFDM) and Flip-OFDM. Both techniques can be used to compensate multipath distortion effects in optical wireless channels. However, ACO-OFDM has been widely studied in the literature, while the performance of Flip-OFDM has never been investigated. In this paper, we conduct the performance analysis of Flip-OFDM and propose additional modification to the original scheme in order to compare the performance of both techniques. Finally, it is shown by simulation that both techniques have the same performance but different hardware complexities. In particular, for slow fading channels, Flip-OFDM offers 50% saving in hardware complexity over ACO-OFDM at the receiver.

I. INTRODUCTION

Optical wireless OFDM must produce real, positive signals for IM/DD transmission, creating bandwidth and power-efficiency trade-offs. The paper analyzes the under-studied Flip-OFDM scheme, proposes modifications, and compares it with ACO-OFDM.

I. INTRODUCTION

  • IM/DD requires real, positive transmitted signals, whereas the equivalent baseband OFDM signal is usually complex.The optical carrier phase cannot carry information in this noncoherent transmission scheme.Source: §I · p. 1
  • Hermitian symmetry yields a real signal at the cost of half the available bandwidth, while high-PAPR OFDM makes the DC bias in DCO-OFDM significant.Large DC bias makes DCO-OFDM optical-power inefficient; lower bias can leave negative samples clipped, causing interference and out-of-band optical power.Source: §I · p. 1
  • ACO-OFDM uses odd subcarriers and clips negative samples without distorting odd-subcarrier symbols, whose amplitude is scaled by half.Prior comparisons suggest better power efficiency than other modulation schemes for optical wireless channels.Source: §I · p. 1
  • Flip-OFDM separates a real bipolar OFDM signal into positive and flipped-negative parts transmitted in two consecutive OFDM symbols.Both transmitted subframes therefore have positive samples.Source: §I · p. 1
  • The paper analyzes Flip-OFDM, proposes improvements, and compares its performance and complexity with ACO-OFDM.The introduction identifies Flip-OFDM as an alternative unipolar OFDM technique that had been largely ignored.Source: §I · p. 1

A. Optical Wireless Channel

The section describes an IM/DD optical wireless link and distinguishes directed from diffused operation. It models the channel between optical transmission and electrical detection, with multipath distortion arising from paths of different lengths.

A. Optical Wireless Channel

  • An infrared emitter sends optical signal x(t), and a photodetector converts the received signal into electrical current y(t).Fig. 1 presents an equivalent baseband channel model for IM/DD optical wireless channels.Source: A. Optical Wireless Channel · p. 1, Figure 1 · p. 2
  • Directed links are modeled with AWGN when line-of-sight dominates, while diffused links without strong line-of-sight use a linear baseband channel model.The model uses channel impulse response h(t) and noise n(t); the received signal is described using convolution.Source: A. Optical Wireless Channel · pp. 1–2
  • Optical power traveling along paths of different lengths contributes to multipath distortion.The channel impulse response can be modeled using a series of Dirac delta functions.Source: 1) Diffused Optical Wireless Channel… · p. 2

1) Diffused Optical Wireless Channel Model:

The diffused-channel discussion attributes multipath dispersion to reflected light and describes channel models for those paths. It also characterizes the dominant noise components as Gaussian and signal-independent under the stated assumptions.

1) Diffused Optical Wireless Channel Model:

  • Multiple reflections and single reflections are the two physical processes identified as producing multipath dispersion.The channel response may be modeled using channel coefficients, a number of paths, and a maximum channel delay.Source: 1) Diffused Optical Wireless Channel… · p. 2
  • The exponential-decay and ceiling-bounce models are used to represent multiple and single optical reflections.The exponential-decay model describes the response due to multiple reflections.Source: 1) Diffused Optical Wireless Channel… · p. 2
  • The multiple-reflection model uses D for rms delay spread and u(t) for the unit step function.The section also introduces a ceiling-bounce channel impulse response model.Source: 1) Diffused Optical Wireless Channel… · p. 2
  • Photon noise and receiver circuit thermal noise are the two dominant noise components described for optical wireless systems.Shot noise is treated as white Gaussian and independent of the transmitted signal; thermal noise is also described as Gaussian and signal-independent.Source: 2) Noise Model · p. 2

2) Noise Model:

Flip-OFDM forms a real bipolar OFDM signal, then sends its positive and inverted negative parts in successive unipolar subframes. The receiver recombines the subframes to recover the bipolar signal for FFT-based detection.

2) Noise Model:

  • The IFFT output is made real by imposing Hermitian symmetry on the subcarriers, which sacrifices half of them.Flip-OFDM uses half of the total subcarriers to carry information.Source: B. Flip-OFDM · p. 2
  • The transmitter sends the positive signal in the first subframe and the inverted negative signal in the second, with cyclic prefixes added to both.The second subframe follows the first after a delay of N + Δ.Source: B. Flip-OFDM · p. 3
  • The original Flip-OFDM proposal compresses the multiplexed samples by a factor of two to match a single bipolar frame’s symbol duration.The compressed guard interval should accommodate the channel’s full delay spread.Source: B. Flip-OFDM · p. 3, Figure 3 · p. 3
  • After cyclic-prefix removal, the receiver subtracts the second subframe’s samples from the first to regenerate the bipolar signal before FFT detection.The two received subframes supply the respective y+(k) and y−(k) samples.Source: B. Flip-OFDM · p. 3

III. COMPARISON OF ACO- AND FLIP-OFDM

The comparison matches the systems’ bandwidth, data rate, and cyclic-prefix length by modifying Flip-OFDM. It evaluates spectral efficiency, electrical-domain SNR, and expected BER under a shared channel assumption.

III. COMPARISON OF ACO- AND FLIP-OFDM

  • The comparison evaluates spectral efficiency, electrical-domain SNR, and expected BER for ACO-OFDM and Flip-OFDM.The systems are first modified so bandwidth, data rate, and cyclic-prefix length match.Source: §III · p. 3
  • Both systems are assumed to use the same optical wireless channel with a given delay spread.The channel is assumed constant over two consecutive OFDM symbols.Source: §III · p. 3
  • The comparison holds bandwidth, data rate, and cyclic-prefix length equal between the two systems.Source: §III · p. 3

A. Modification of Flip-OFDM

The authors modify Flip-OFDM by leaving its two OFDM frames uncompressed. This makes the comparison with ACO-OFDM use matching cyclic-prefix lengths and, with equal FFT/IFFT sizes and sampling rates, matching bandwidth.

A. Modification of Flip-OFDM

B. Comparison of key parameters

ACO-OFDM and Flip-OFDM have the same spectral efficiency and electrical-domain BER, despite using transmitted energy differently. In Flip-OFDM, recombination doubles both equivalent electrical energy per symbol and noise power.

B. Comparison of key parameters

  • Both schemes have the same spectral efficiency: Flip-OFDM uses twice as many samples to transmit twice as many information symbols.ACO-OFDM’s spectral efficiency is one fourth of that of a typical OFDM system with complex signals, i.e., (1/4 × log2(M)).Source: 1) Spectral Efficiency · pp. 3–4

2) Electrical domain SNR and expected BER performance:

  • ACO-OFDM shifts half of its transmitted energy to even subcarriers as clipping noise, while Flip-OFDM recombines energy spread across two OFDM symbols.The recombined Flip-OFDM symbol has doubled noise power even though its transmitted symbol energy is not wasted.Source: 2) Electrical domain SNR and expected… · p. 4
  • Both schemes have the same electrical-domain BER because Flip-OFDM’s recombination doubles both equivalent electrical energy per symbol and noise power.Electrical-domain SNR is defined as E[xt2(k)]/σ2 and is directly related to expected BER.Source: 2) Electrical domain SNR and expected… · p. 4

IV. SIMULATION RESULTS

This section compares the BER performance and hardware computation complexity of ACO-OFDM and Flip-OFDM.

IV. SIMULATION RESULTS

  • The section compares the BER performance of ACO-OFDM and Flip-OFDM.Source: §IV · p. 4
  • The section also compares the hardware computation complexity of both schemes.Source: §IV · p. 4
  • The comparison covers both performance and hardware computation complexity.Source: §IV · p. 4

A. BER

The BER comparison covers direct AWGN and diffused optical wireless channels, and the reported electrical BER is the same for ACO-OFDM and Flip-OFDM in both.

A. BER

  • ACO-OFDM and Flip-OFDM have the same electrical BER in both direct AWGN and diffused optical wireless channels.Fig. 6 presents the BER comparison; the diffused-mode optical wireless channel is normalized so that ∥h(t)∥2 = 1.Source: Figure 6 · p. 4, A. BER · p. 4
  • The Fig. 6 comparison uses channel coefficients generated half from the Ceiling-Bounce Model and half from the Exponential-Decay Model.The coefficients are uniformly distributed from 0 to he(t) or hc(t, a).Source: Figure 6 · p. 4

B. Complexity of TX and RX operations

Under slow fading, Flip-OFDM saves 50% in receiver complexity while matching ACO-OFDM in BER performance and spectral efficiency in the electrical domain. Complexity counts FFT/IFFT operations; with optimized ACO-OFDM IFFT, transmitter computation complexity is equal for both schemes.

B. Complexity of TX and RX operations

V. CONCLUSION

  • Under slow fading, Flip-OFDM saves 50% in receiver complexity while matching ACO-OFDM in BER performance and spectral efficiency in the electrical domain.Source: §V · p. 5
Loading 1111.5682v1…