<?xml-model href='http://www.tei-c.org/release/xml/tei/custom/schema/relaxng/tei_all.rng' schematypens='http://relaxng.org/ns/structure/1.0'?><TEI xmlns="http://www.tei-c.org/ns/1.0">
	<teiHeader>
		<fileDesc>
			<titleStmt><title level='a'>J-Band Communication System considering the Near-to-far field Transitions - An Experimental Approach</title></titleStmt>
			<publicationStmt>
				<publisher>IEEE</publisher>
				<date>10/26/2025</date>
			</publicationStmt>
			<sourceDesc>
				<bibl> 
					<idno type="par_id">10705586</idno>
					<idno type="doi">10.1109/IEEECONF67917.2025.11443935</idno>
					
					<author>Priyangshu Sen</author><author>Justin Osmond</author><author>Arjuna Madanayake</author><author>Arjun Singh</author>
				</bibl>
			</sourceDesc>
		</fileDesc>
		<profileDesc>
			<abstract><ab><![CDATA[Not Available]]></ab></abstract>
		</profileDesc>
	</teiHeader>
	<text><body xmlns="http://www.tei-c.org/ns/1.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xlink="http://www.w3.org/1999/xlink">
<div xmlns="http://www.tei-c.org/ns/1.0"><head>I. INTRODUCTION</head><p>The terahertz (0.1-10 THz) band has gained acceptance as a likely candidate in sixth-generation (6G) and future wireless communication standards. The massive available bandwidth can facilitate emerging applications through <ref type="bibr">[1]</ref>, <ref type="bibr">[2]</ref>: i) exceptionally high data rates, ii) high-resolution sensing capabilities, and iii) ultra-secure communication networks. This has motivated research progress originating from device technology to networking and link layer protocols <ref type="bibr">[3]</ref>. In recent years, point-to-point multi-km THz communications have been demonstrated <ref type="bibr">[4]</ref>, <ref type="bibr">[5]</ref>. At the same time, it has been observed that the high antenna gains required due to propagation losses in the THz regime make THz near-field communications likely in mobile and indoor communication scenarios.</p><p>As such, several studies have been recently undertaken <ref type="bibr">[6]</ref>- <ref type="bibr">[10]</ref>, with the aim to answer -what exactly changes in the THz near-field? First, the Friis path loss equation does not hold. A simple example is that if the "path loss" is found to be less than the combined gain of the transmitter and receiver antenna, we anticipate more power at the receiver than transmitted -an obvious fallacy <ref type="bibr">[6]</ref>! Second, to accurately quantize and characterize THz near-field propagation, it is important to account for the impact of the antenna on the channel profile. In fact, the studies conducted so far have primarily focused on fixed antenna setups, overlooking a crucial aspect of antenna effects and near-field operation. In real-world THz application scenarios, it is highly likely that we will be operating in the near field, and nodes may transition between the near and far fields of the antenna. The importance of this understanding is highlighted in [?], where it is mathematically shown that mobile THz communication nodes need to operate within the near field. The testbed developed to date has primarily concentrated on the communication and channel sounding aspects, while overlooking the near-field propagation mechanisms and their experimental validation. Although some studies have acknowledged the influence of antenna effects on channel behavior, they have not provided a comprehensive demonstration of this effect. Specifically, the near-field THz channel cannot be designed agnostic of the actual antenna(s) <ref type="bibr">[11]</ref> on the transmitter and receiver. In addition, new propagation mechanisms, such as wavefront manipulation and beam shaping to realize Focusing, Bessel and Airy beams <ref type="bibr">[8]</ref>- <ref type="bibr">[10]</ref> become possible, motivating an investigation on the limits of conventional beamforming.</p><p>In this paper, we present practical measurement results and characterization of indoor THz links at 300 GHz, where the near-field and antenna effects on the channel are comprehensively accounted for. In this task, we first present the advanced communication electronics and sensing (ACES) testbed, designed to capture the near-field and antenna effects on THz communication and propagation mechanisms. We provide a near-field specific link budget with an approach different from the Friis path loss, as well as channel modeling results in typical indoor setups accounting for multipath affects. Our results reveal the significant impact of reflections with larger near-field as well as the possibility of asymmetrical up/down-link channel capacities in THz communication links, motivating further investigation into physical and link layer design for THz communication systems.</p><p>The rest of the paper is organized as follows. In Section II, we describe the capabilities of the ACES near-field testbed in terms of hardware and the signal processing backend. Further, in Section III, we experimentally characterize the testbed and present the corresponding results in Sec. IV. Finally, in Section V, we conclude our work.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>II. ACES TESTBED</head><p>The ACES testbed is designed for experimental applications for sub-THz and THz band communications. It operates at 300 GHz in compliance with the only standard for THz wireless links, the IEEE 802.15.3-2023 (previously, IEEE 802.15.3d) <ref type="bibr">[12]</ref>. The testbed has two main components: the hardware system and the signal processing backend.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>A. Hardware System</head><p>The hardware system is divided into the transmitter and the receiver section (Fig. <ref type="figure">1 (a)</ref>). At the transmitter, baseband(BB)/intermediate frequency(IF) signals are generated by an arbitrary waveform generator (AWG) from Keysight Technologies, at a sampling rate of 65 GSa/s (M8195A, analog bandwidth of 25 GHz). The output from the AWG is mixed with a highly stable local oscillator (LO) signal with low phase noise (-100 dBc/Hz at a 10 Hz offset), generated from the 865-40-LN signal generator by BNC. Thereafter, the mixed signal is converted to the RF domain. The integrated mixer and up-converter is supplied by VDI with a double sideband conversion loss of 12 dB and a &#215;12 multiplier. The RF is generated at the J-band (220-330 GHz). The up-converted signal is then amplified using a 26 dB amplifier and transmitted utilizing different antennas, with gains ranging from 26-46 dBi. The maximum achievable transmit power at the radio frequency (RF) port is 8.75 dBm. At the receiver, the signal is down-converted using a similar VDI multiplier and mixer chain. Further, the down-converted signal is amplified using a low noise amplifier (LNA) at IF with 12 dB of gain. The down-converted signal is then sampled, captured, and stored using a digital storage oscilloscope (DSO) by Tektronix (MS064B) with a sampling rate of 50 GSa/s and an analog bandwidth of 4 GHz. The signal is then processed using the signal processing backend.</p><p>The near-field range, also called the Fruanhoffer distance d F , is given as d f = 2D 2 /&#955;, where &#955; is the signal wavelength and D is the antenna aperture. With interchangeable antennas, we can analyze near-field effects, as we outline in Section IV.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>B. Signal Processing Backends</head><p>The MATLAB based signal processing is divided into two sections: i) communication; and ii) channel sounding.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>1) Communication Backend</head><p>The communication backend consists of the transmitter and receiver (Fig. <ref type="figure">1 (b)</ref>). At the transmitter, bits are generated and encapsulated within frames, which are then modulated into symbols. Each frame consists of a header, a training sequence, and a data sequence. The header utilizes a maximum length sequence (MLS) <ref type="bibr">[13]</ref> to ensure high auto-correlation and low cross-correlation to detect the start of the frame in a low SNR scenario. The training sequence is employed to estimate and equalize the channel. The frame is modulated and pulse-shaped using customizable single and multi-carrier modulation schemes. To address the frequency-selective response of the hardware, a pre-equalization filter is applied.</p><p>At the receiver, a bandpass filter with a high roll-off factor is utilized to limit out-of-band noise. Frame synchronization is achieved by correlating with the header sequence. To mitigate the effects of multipath and the frequency-selective characteristics of the channel, a post-equalization filter based on maximum mean square error (MMSE) statistics is applied. Finally, a correlation-type detector using the maximum likelihood (ML) criterion is used to detect the received signal.</p><p>A constellation diagram for 8-PSK at 2 Gbps at a distance of 2 m and 300 GHz carrier with 25 dBi antennas at TX and RX is shown in Fig. <ref type="figure">1d</ref>). This communication scheme achieves a low bit error rate (zero over 30000 bits).</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>2) Channel Sounding Backend</head><p>Our tailored sliding correlator-type channel sounder can accurately obtain the multipath profile of ultra-broadband THz channels (Fig. <ref type="figure">1 (c)</ref>). At the transmitter, a 4095-length MLS is utilized, modulated using binary phase shift keying (BPSK), and pulse-shaped with a root-raised cosine pulse that has a high roll-off factor. At the receiver, the received signal is calibrated to eliminate frequency-selective responses of the hardware. Thereafter, a dual phase-locked loop (PLL) is used to synchronize with the transmit carrier and eliminate any carrier frequency and phase offset. The signal is then correlated with the locally generated 4095-MLS signal to extract the channel impulse response, averaged over 16 repetitions to reduce noise. The received multipath power profile is normalized by the pulse energy and MLS length (4095). Peaks and delays are measured using a global convergence method based on the channel sounder's dynamic range. The channel sounder is capable of measuring multipath components with 0.25 ns time resolution (as 4 GHz bandwidth is utilized). It can estimate the multipath channel response up to 142 dB of path loss, with a dynamic range of 60 dB.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>III. TESTBED CHARACTERIZATION</head><p>The characterization of the testbed is crucial for understanding its limitations as well as capabilities and designing the end-to-end communication system.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>1) Frequency Selectivity</head><p>We conduct frequency characterization of the testbed by measuring the received power at a distance of two meters across the operational bandwidth of the up-converter, i.e., 220 to 330 GHz in 10 GHz steps. The results of this characterization are illustrated in Fig. <ref type="figure">2a</ref>). We observe a highly frequency-selective response. Thus, we utilize a post-equalization filter at the receiver, which helps to restrict further noise amplification while post-processing the signal and maintaining a high SNR.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>2) Amplitude Noise Characterization</head><p>The amplitude noise refers to the additive noise incorporated by the receiver, which encompasses both external  (environmental and space noise) and internal noise (receiver components, including the mixer and LNA <ref type="bibr">[14]</ref>). In our system, due to the absence of an LNA at RF, we directly input the RF signal into the mixer. As a result, the noise floor of the receiver is elevated, with an N 0 value of 2.7 &#215; 10 -19 W/Hz. The approximated mean and variance of the noise are 0.07mV and 0.00108&#181;W , respectively, for 4 GHz of bandwidth, as shown in Fig. <ref type="figure">2b</ref>). Additionally, the noise follows Gaussian statistics, justifying the use of a maximum likelihood (ML) detector in the demodulator.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>3) Phase Noise Characterization</head><p>PN is primarily generated in the LO signal due to jitter <ref type="bibr">[15]</ref>. Additionally, the noise level increases due to the use of a multiplier chain in the transceiver module. In the frequency domain, this phenomenon is typically expressed as single-sideband (SSB) phase noise, which represents power leakage per hertz of bandwidth at various frequency offsets from the carrier component. The phase noise statistics are presented in Fig. <ref type="figure">2c</ref>). We observe a low phase noise of -100 dBc/Hz at 10 MHz, even with a &#215;12 multiplier chain due to the stable LOs. Based on the phase noise, we restrict the maximum frame duration to 100 ns, as most of the energy related to PN is concentrated within a 10 MHz frequency offset.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>IV. EXPERIMENTAL VALIDATION 1) Link Budget</head><p>Link budget analysis is crucial for correctly estimating the received power and maintaining the required link margin for effective communication. In dB, the received power P rx , is given as</p><p>where, P tx represents the transmitted power, G LN A refers to the gain of the LNA at the receiver. The term L pr indicates propagation loss, while L abs represents absorption loss <ref type="bibr">[16]</ref>, L mix is the conversion loss at the receiver, and L misc accounts for miscellaneous losses associated with cables and connectors <ref type="bibr">[17]</ref>.</p><p>The propagation loss, L pr , in near-field differs from the Friis path loss equation <ref type="bibr">[18]</ref>. A more accurate estimation is to directly account for the power at the receiver for a given transmitter-receiver setup. This follows from knowing that the signal power and the electric field of the signal are interrelated, and utilizing electromagentic theory for exact calculations, as in optics. If we assume a circular antenna surface (as in our setups), we can utilize the Gaussian beam approximation <ref type="bibr">[19]</ref>.</p><p>It has been shown that Gaussian beams have a power density that can be calculated by knowing that the signal amplitude spreads out over a beam waist w z :</p><p>where w tx is the radius of the transmitter and z R = &#960;w 2 0 /&#955; is the Rayleigh range, and is a measure of how fast the beam spreads out. The resultant power density at a communication distance z is accordingly found as:</p><p>and the received power P rx,GB , captured by a receiver antenna with radius w rx and effective area &#960;w 2 rx , is given by</p><p>The path loss L pr is then found through the ratio of P rx,GB Ptx and converting to dB. We observe from (4) that if we swap the RX and TX, we observe a different path loss for unequal antennas. This is validated in Fig. <ref type="figure">3</ref>, where the received power with a 46 dBi antenna at the transmitter and a 26 dBi antenna at the receiver is significantly different than with a 26 dBi Tx and a 46 dBi RX. This leads to an asymmetrical capacity for the uplink and downlink. Further, observing the experimental and theoretical power values in Fig. <ref type="figure">3</ref> and comparing these to the Friis path loss model, it is evident that our path loss model aligns more accurately with the experimental results. </p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>2) Channel Sounding</head><p>An instance of a multipath profile at 300 GHz for a 360-degree angle captured in an indoor corridor setup with a 2 m communication distance is shown in Fig. <ref type="figure">4</ref>. It is observable that although the channel is parsed and the link setup across 360 degrees is not possible, there are multipath components beyond line of sight (LoS) that could be utilized for reliable link establishment. An efficient and innovative link-searching algorithm is essential, especially given the use of high-gain antennas with a pencil-thin beam for THz communication. Furthermore, the system may operate in the near field region of the antenna, which necessitates precise alignment.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>ACKNOWLEDGMENT</head><p>This work is partially supported by the U.S. National Science Foundation under grants NSF IIS-2509392, NSF CNS-2451446, and the WINGS Center at SUNY Polytechnic Institute.</p></div>
<div xmlns="http://www.tei-c.org/ns/1.0"><head>V. CONCLUSION</head><p>This paper presents novel measurements for indoor THz communications, where the near-field and channel sounding both demonstrate the significant impact of antenna configurations in producing asymmetrical up/down-link communication channels. The work is verified through a specific testbed designed to enable experimentation across these complex setups, effectively paving the way for mobile THz communication investigations and other emerging application use cases. The near-field effects, phase noise, frequency selectivity at the IEEE standard compliant 300 GHz frequency all reveal the immense potential of the THz band if the underlying phenomena are correctly accounted for. Our future work is aimed at extending our work to near-field MIMO, as well as utilizing ML based algorithms for frame design and error correction. We also intend to expand the beamshaping capabilites of our testbed by exploring Bessel and focusing beams through custom lenses.</p></div><note xmlns="http://www.tei-c.org/ns/1.0" place="foot" n="2025" xml:id="foot_0"><p>59th Asilomar Conference on Signals, Systems, and Computers | 979-8-3315-8745-1/25/$31.00 &#169;2025 IEEE | DOI: 10.1109/IEEECONF67917.2025.11443935 Authorized licensed use limited to: Northeastern University. Downloaded on July 26,2026 at 16:03:03 UTC from IEEE Xplore. Restrictions apply.</p></note>
			<note xmlns="http://www.tei-c.org/ns/1.0" place="foot" xml:id="foot_1"><p>Authorized licensed use limited to: Northeastern University. Downloaded on July 26,2026 at 16:03:03 UTC from IEEE Xplore. Restrictions apply.</p></note>
		</body>
		</text>
</TEI>
