Research

1: High-Flux Optical Link Design and Hardware Implementation

We have successfully validated a high-performance system for simultaneous ranging and communication, confirming the viability of our integrated design. The core of this success lies in our novel hybrid tag architecture, which resolves the fundamental trade-off between ranging accuracy and communication signal strength. Through systematic optimization of the light source, tag, and receiver, we identified an optimal hardware configuration built entirely from cost-effective, commercially available components, demonstrating the practicality of our solution. 

(a) Light Source Design

To maximize the communication link's signal-to-noise ratio (SNR), our final light source array employs a non-uniform, center-dense LED layout that leverages the higher optical efficiency of the central region. The key challenge of this approach, localized overheating, was addressed by using an aluminum-core PCB (MCPCB). This solution was validated by experiments on a 30-LED test board, which confirmed the MCPCB's superior heat-spreading ability by showing a temperature differential of only 3°C between the dense center and the sparse edge. While this confirmed excellent heat distribution, the test also established that the board's overall operating temperature neared the 80°C limit for the LEDs. This led to the critical conclusion that a large heat sink is a mandatory component for the long-term reliability of the final 216-LED array. Our final design, as depicted in Figure 2, incorporates these findings, consisting of 216 LEDs in six parallel groups with spacing that increases linearly from 5 mm at the center. 

(b) Design of the Hybrid Retro-reflective Tag

Our investigation aimed to identify the optimal components for both an unmodulated retro-reflector (URR) and a modulated retro-reflector (MRR).

For the unmodulated case, we tested various commercial retro-reflectors and concluded that the Total Internal Reflection (TIR) corner cube is the superior choice, consistently providing the highest signal return efficiency. While a flexible 3M reflective sheet showed the widest response angle, the TIR cube offered the best performance for a fixed area.

However, for the modulated tag, integrating a Liquid Crystal (LC) device with a standard TIR cube introduces a critical challenge: a severe polarization mismatch that drastically attenuates the return signal. To solve this, we systematically tested combinations of five retro-reflectors and four different LC devices.

Our results revealed a complete reversal in performance rankings. The combination of a hollow, aluminum (Al) coated corner cube paired with a 5V-driven Twisted Nematic (TN) liquid crystal shutter proved to be the optimal configuration. This pairing effectively overcomes the polarization issue and achieves the highest overall system efficiency. Other combinations were less effective; notably, the Polymer-Dispersed Liquid Crystal (PDLC) system was found to be unfeasible due to excessive optical losses.

(c) Optical Reader Design and Selection 

Our final conclusion is that an integrated, commercial photodetector is the unequivocally superior and more reliable choice for the optical reader.

We initially developed and tested a custom array of four parallel photodiodes. While theoretically promising for higher signal-to-noise ratio, this custom design failed in practice due to critical engineering flaws. The unshielded, hand-wired array suffered from severe electromagnetic interference (EMI) and crosstalk from the light source driver. Furthermore, the high capacitance from paralleling the diodes drastically limited the detector's bandwidth, making it unsuitable for high-speed signals.

The performance gap was quantified through direct noise measurements. The commercial detector exhibited a noise floor of -97 dBV, which was over 25 dB lower than our best custom configuration (-71.6 dBV). We further confirmed the critical role of shielding in our custom build, as simply shortening the unshielded wires from 12 cm to 2 cm reduced the noise level by 4.4 dB. These results provide conclusive evidence that professional circuit layout and electromagnetic shielding are paramount, making the commercial detector the only viable option to achieve the high-sensitivity performance required by our system.

2: PHY-Layer Scheme for Multi-Tag Communication and Ranging 

To address the challenge of multi-tag concurrent transmission, traditional solutions typically involve designing a scheduling-based MAC protocol (e.g., TDMA) to temporally separate signals and avoid collisions. This project, however, introduces an innovative physical-layer scheme where we allow signals from different tags to superimpose at the receiver and then successfully separate them using advanced signal processing techniques.

The core principle of this scheme is as follows:

1. Tag Differentiation and Ranging via Time Delay (Sensing): When the optical signal simultaneously illuminates multiple tags at different physical locations, their return signals arrive at the receiver with unique time delays corresponding to their unique distances. After these signals superimpose, we perform a cross-correlation operation on the composite received signal. This operation produces multiple discrete peaks along the time axis. The position of each peak precisely corresponds to the time delay of a specific signal, from which the exact distance to the tag can be calculated. In this manner, we can identify and locate all tags with a single operation.

2. Communication Demodulation via Peak Amplitude (Communication): Once the peak positions for each tag are identified and locked, we continuously monitor the amplitude of a specific peak over time. Since the tags use On-Off Keying (OOK) for modulation, the peak's amplitude will be high when the tag reflects light (transmitting a '1') and low or absent when it does not (transmitting a '0'). Therefore, the real-time fluctuation of the peak's amplitude perfectly recovers the data stream sent by the tag.

Through this design, the peak's position carries spatial information (for ranging and differentiation), while the peak's amplitude carries the communication data. This achieves an elegant separation of multi-user signals and integrated sensing and communication without the need for traditional MAC scheduling.

3: Analysis of Sensing-Communication Performance Relationship and Resource Allocation Strategy 

Our research leads to a core conclusion: under the current hardware constraints, the bottleneck for communication performance is the low modulation rate (several hundred Hertz) of the Liquid Crystal (LC). Through experimental comparison, we found that at this rate, the impact of On-Off Keying (OOK) modulation on the ranging accuracy of the high-bandwidth OFDM signal is negligible. This indicates that the sensing and communication tasks operate in a virtually "conflict-free" manner in the current system, achieving a natural decoupling in resource usage. 

4:  Simulation Conclusions for the ISAC System 

Our simulation work was structured to address the core challenge of achieving high-resolution positioning despite the limited bandwidth of commercial LEDs. The low bandwidth of our LED panel (4 MHz) requires a very high oversampling rate to achieve the 2.5 GHz effective sampling needed for centimeter-level accuracy, which presents challenges in peak detection. Our simulations systematically investigated this trade-off.

1. Feasibility with Real-World Hardware: The first simulation confirmed that centimeter-level positioning is feasible even with our measured 4 MHz LED bandwidth. However, this is limited to a single large delay region, where location is inferred from the correlation peak's amplitude rather than its position. We also found that accuracy is non-uniform, degrading at the center (due to 1D-to-2D mapping geometry) and at the far edges (due to low SNR).

2. Dominance of Oversampling Rate: The second study investigated the trade-off between SNR and oversampling rate (OSR). The key conclusion is that the oversampling rate is the dominant factor determining peak detection accuracy, far more than SNR. Across a wide SNR range (10-40 dB), accuracy remained stable for a given OSR, proving the system is robust to varying signal strength but highly dependent on the effective bandwidth.

3. Success of the Multi-Robot Hybrid Tag System: The final simulation validated the complete system concept with multiple robots. The main conclusions are:

   • The hybrid URR-MRR tag structure successfully enables simultaneous localization and communication for multiple robots.

   • The system can reliably separate signals from two robots, even when they are in adjacent delay regions.

   • Successive Interference Cancellation (SIC) is critical for recovering weak signals when robots are far apart, significantly improving both ranging and communication performance (BER). However, we also found that when robots are in adjacent regions, SIC is not always beneficial, as the overlapping correlation peaks can reinforce each other, making them easier to detect without cancellation.