betaFeedback

WorksComputer Science

System-Level Evaluation of Uplink Access Schemes for Ambient IoT

Abstract

Ambient Internet of Things (A-IoT) devices operate under stringent energy constraints, relying on harvested power in the μμW-to-mW range and limited energy storage. The A-IoT air interface for active devices is still under development; we therefore use Narrowband Internet of Things (NB-IoT) as an implementable reference for evaluating autonomous Uplink (UL) access under harvested energy. Under these conditions, the standard NB-IoT contention-based Random Access (RA) procedure incurs substantial per-packet overhead, making UL access in dense A-IoT networks increasingly energy-expensive. Context-retaining UL access schemes reduce this overhead, but their energy-latency trade-offs are insufficiently studied in literature under realistic harvesting and protocol constraints. To address this gap, we extend the LENA-NB ns-3 module with solar energy harvesting, supercapacitor storage, and a hysteretic brown-out mechanism, and propose Hybrid Scheduling Request (HSR), a hybrid UL access scheme combining reserved request occasions with opportunistic contention. We compare RA, Scheduling Request (SR), HSR and an Idealized Fast Uplink Grant (FUG) serving as a performance bound. Results show that HSR reduces mean packet delay by up to 69% and SR reduces energy per delivered bit by up to 24%, both relative to RA.

Introduction

The Third Generation Partnership Project (3GPP) is standardizing Ambient Internet of Things (A-IoT) as a new class of cellular Internet of Things (IoT) devices targeting battery-less operation through ambient energy harvesting [1]. Release 19 specifies an air interface for passive devices that backscatter a carrier provided by the network [2, 3]. Release 20 extends the framework to active devices, including types 2b and C, which generate their own transmissions and therefore do not require an externally supplied carrier. The air interface for these active device types remains unspecified in the current Release 20 work item [4]. We therefore use Narrowband Internet of Things (NB-IoT), an already deployed low-power cellular IoT technology, as an implementable reference. Unlike conventional battery-powered NB-IoT User Equipments (UEs), A-IoT devices operate under extremely constrained energy budgets of microwatts to milliwatts. The harvested energy is commonly buffered in supercapacitors, which have lower energy density and higher self-discharge than batteries, making device operation dependent on instantaneous energy availability. Furthermore, their traffic is inherently sparse, consisting of short and infrequent Uplink (UL) reports rather than sustained data flows. Under these conditions, control-plane signaling can dominate the overall energy consumption, while insufficient harvested energy may trigger brown-out events that interrupt both communication and application execution [5].

The standard NB-IoT UL access procedure introduces substantial per-packet overhead through the four-step contention-based Random Access (RA) process, requiring preamble transmission, random-access response, contention resolution, and UL resource allocation before application data delivery [6]. Context-retaining schemes, such as Semi-Persistent Scheduling (SPS), Preconfigured Uplink Resources (PUR), and Group Wake-Up Signal (GWUS) [7, 8], as well as predictive Fast Uplink Grant (FUG) scheduling [9], mitigate this repeated overhead by allowing lightweight data requests or proactive resource allocation. However, they introduce additional listening, state-retention, and resource-partitioning costs. Moreover, reserved request opportunities must be shared among devices, causing latency to increase with the device density.

The suitability of context retention under active A-IoT operation depends jointly on traffic characteristics, harvesting dynamics, and network load, and has not been quantified. Similar access and context-management choices are still open in the standardization process. For types 2b and C devices, current agreements allow either contention-based access carrying data or a preamble-only request transmitted on resources shared by multiple devices [10]. Whether a device may retain its temporary identifier and thereby avoid repeated random access also remains undecided [11]. Existing work either assumes periodic traffic (SPS, PUR), accurately predictable traffic (FUG), or unconstrained energy availability. Hybrid approaches combining reserved and opportunistic request opportunities remain unexplored.

To address this gap, we extended the LENA-NB ns-3 framework [12] with a harvest-aware A-IoT energy front-end comprising a solar energy harvester, a supercapacitor-based storage model, and a hysteretic brown-out mechanism that couples energy availability with both traffic generation and radio operation. Building upon this framework, we implement: (i) context-retaining UL access through dedicated Scheduling Requests (SRs) on reserved Narrowband Physical Random Access Channel (NPRACH) resources and (ii) propose Hybrid Scheduling Request (HSR), a hybrid scheme that combines reserved requests with energy-aware opportunistic contention. SR represents preamble-only requesting when request resources are dedicated, HSR extends this approach to request resources shared among devices, and RA with Early Data Transmission (EDT) represents contention-based access that carries data within the access procedure [10, 11]. This enables the first system-level comparison of RA, SR, and HSR, under identical harvesting, traffic, and propagation conditions, with an Idealized FUG configuration bounding the achievable latency and energy. Sections II–III present the access schemes and related work, Sections IV–VI detail the system model, simulation platform, and results, and Section VII concludes the paper.

Uplink Access Schemes

This section presents the UL channel access schemes considered in this work: RA, FUG, SR, and HSR. The signaling procedures of the schemes are illustrated in Figure 1.

UL access schemes: RA with EDT, SR, and FUG

Figure 1. UL access schemes: RA with EDT (left), SR (center), and FUG (right).

Random access (RA) with EDT

Under RA, each UL transmission is preceded by the standard two- to four-step contention-based access procedure defined in 3GPP TS 36.321. A UE returning from Power Saving Mode (PSM) or Extended Discontinuous Reception (eDRX) first transmits an MSG1 preamble over the configured NPRACH resources. After successful detection, the Base Station (BS) responds with MSG2 containing a temporary identifier and an UL grant for MSG3 transmission, which carries the Radio Resource Control (RRC) connection request and, in an EDT configuration, UL payload. The procedure is completed through contention resolution (MSG4). If multiple UEs select the same preamble, the affected devices may perform back-off and retransmission up to the configured attempt limit, after which the packet is discarded. Following transmission completion, the UE returns to eDRX for duration T3324T_{3324} before transitioning back to PSM until the next reporting event or expiration of timer T3412T_{3412}.

From a resource-allocation perspective, each RA attempt consumes a sequence of NPRACH, Narrowband Physical Downlink Control Channel (NPDCCH), and Narrowband Physical Uplink Shared Channel (NPUSCH) resources whose availability is bounded by the cell broadcast configuration. NPRACH opportunities are configured with periodicity TNPRACH∈[40 ms,2.56 s]T_{\mathrm{NPRACH}} \in[40\ \mathrm{ms}, 2.56\ \mathrm{s}], with each occasion reserving PscP_{\mathrm{sc}} contiguous subcarriers for RA transmission. A single-tone Format-0 preamble occupies Nrep⋅5.6 msN_{\mathrm{rep}} \cdot5.6\ \mathrm{ms}, where the repetition factor Nrep∈{1,…,128}N_{\mathrm{rep}} \in\{1,\ldots,128\} is selected according to the coverage enhancement level. Following MSG1 reception, the BS returns MSG2 within the configured random-access response window WRARW_{\mathrm{RAR}}, while the uplink scheduling delay for MSG3 is determined by the parameter k2∈{12,16,32,64} msk_2 \in\{12,16,32,64\}\ \mathrm{ms}. The resulting access delay can therefore be approximated as

τRA≤TNPRACH+WRAR+k2+TNPUSCH+TCR,\tau_{\mathrm{RA}} \leq T_{\mathrm{NPRACH}}+W_{\mathrm{RAR}}+k_2+T_{\mathrm{NPUSCH}}+T_{\mathrm{CR}},

where TNPUSCHT_{\mathrm{NPUSCH}} denotes the MSG3 transmission duration and TCRT_{\mathrm{CR}} represents the contention-resolution turnaround time.

FUG requires the full RA procedure only during the initial UL transmission of each UE. Once attached, both the UE and the BS retain the transmission context established during the initial RA procedure, including the Cell Radio Network Temporary Identifier (C-RNTI), timing-advance information, and all state information required to resume scheduled UL transmissions without re-initiating RA across sleep cycles. For stationary UEs, where the propagation delay remains effectively constant, the stored context remains valid across power-saving periods (i.e., PSM and eDRX), and the BS proactively schedules UL transmissions via Downlink Control Information (DCI) 0 grants over the User-Specific Search Space (USS), whose periodicity is TUSS=Rmax⁡⋅GT_{\mathrm{USS}} = R_{\max} \cdot G sub-frames, where Rmax⁡R_{\max} denotes the maximum number of NPDCCH repetitions and G∈{1.5,2,4,…,64}G \in\{1.5,2,4,\ldots,64\} the scheduling-gap factor. After decoding the grant, the UE transmits the corresponding NPUSCH burst following a scheduling delay k0∈{8,16,32,64} msk_0 \in\{8,16,32,64\}\ \mathrm{ms}, so the uplink delay is bounded as

τFUG≤TUSS+k0+TNPUSCH.\tau_{\mathrm{FUG}} \leq T_{\mathrm{USS}}+k_0+T_{\mathrm{NPUSCH}}.

What remains is how the BS decides when to push a grant: an idealized variant, used here as a performance bound, learns of each buffer arrival instantly at no signaling cost, whereas a deployable variant must predict each UE’s reporting instants from its observed traffic history. The idealized variant therefore bounds both the achievable latency and the achievable energy of request-based access.

Scheduling Request (SR) and Hybrid SR (HSR)

As in FUG, under SR both the UE and the BS retain the connection context across periods of inactivity and PSM intervals once the initial UL transmission has been completed. While SR transmission on dedicated NPRACH resources is standardized for connected UEs in Rel-15, its use here as a resume trigger from PSM with retained context extends the standard; a functionally equivalent realization is introduced in Rel-16 through PUR. Following the initial connection setup, the BS reserves a specific NPRACH occasion and sub-carrier for each UE and associates these resources with its retained connection context. The resource reservation is valid for a period of TresT_{\mathrm{res}} and is renewed upon each access and during the periodic Tracking Area Update (TAU) procedure. If the reservation expires, the UE loses its reserved resources and must re-establish the connection through RA. When data arrives, the UE wakes, re-synchronizes via NPSS, NSSS and MIB, and requests an UL grant by transmitting a preamble during its pre-assigned NPRACH occasion on its reserved sub-carrier. Upon detecting the preamble, the BS allocates an uplink grant via DCI 0. Since only a fraction of the NPRACH sub-carriers can be reserved, the available resource elements are time-multiplexed among the UEs in a round-robin fashion. Consequently, a given UE’s reserved occasion recurs only every TSR=Tbase⋅⌈N/Nres⌉T_{\mathrm{SR}} = T_{\mathrm{base}} \cdot\lceil N/N_{\mathrm{res}} \rceil, where TbaseT_{\mathrm{base}} denotes the base NPRACH occasion period, NN is the number of UEs, and NresN_{\mathrm{res}} is the number of reserved sub-carriers. With Nres=6N_{\mathrm{res}}=6 and Tbase=80 msT_{\mathrm{base}}=80\ \mathrm{ms}, TSRT_{\mathrm{SR}} grows linearly with NN, exceeding one second beyond N≈75N \approx75; at scale, request latency rather than data transmission therefore dominates the overall uplink delay.

To mitigate this request latency, we propose a hybrid scheme that combines dedicated and contention-based request opportunities. Rather than waiting for its reserved NPRACH occasion, a UE with pending data may transmit a preamble on a randomly selected non-reserved sub-carrier during each base NPRACH occasion, provided that sufficient energy is available and the remaining time before its assigned NPRACH occasion is sufficient to complete the RA procedure. If the preamble transmission succeeds, the UE proceeds with the remaining RA steps and transmits its pending data. Otherwise, it retries at the next base NPRACH occasion while the timing constraint remains satisfied. Consequently, the request latency is bounded by approximately TbaseT_{\mathrm{base}} and TSRT_{\mathrm{SR}}, corresponding to the opportunistic and reserved request opportunities, respectively.

The energy cost of contention-based RA access in NB-IoT has been studied both analytically and experimentally. Prior work has characterized NB-IoT energy consumption across PSM, eDRX, and connected states [13, 14], while measurements on commercial hardware identify NPRACH transmission as a dominant contributor to the per-packet energy budget for short UL payloads [6]. Other studies analyse the NPRACH–NPUSCH resource trade-off and its impact on collision-limited operation [15]. However, these works assume an unconstrained energy supply and therefore do not consider A-IoT devices.

To reduce repeated access overhead, 3GPP introduced persistent UL schemes such as SPS, PUR, and GWUS [7, 16]. In parallel, the FUG scheme was proposed [9], followed by predictive and learning-based schedulers for massive Machine Type Communications (MTC), including multi-armed-bandit allocation [17] and learning-based traffic-prediction and activation models [18]. Closer to NB-IoT, a persistent periodic scheduler exploiting power-saving and suspend/resume procedures has been proposed [19], while grant-free and FUG access have been analytically benchmarked against RA [20]. However, existing learning-based approaches consider simplified system models with a focus on the MAC layer, whereas NB-IoT-specific studies rely on fixed periodic heuristics [19] or analytical benchmarking [20]. Consequently, the performance trade-offs of the UL access schemes under energy-harvesting constraints remain unexplored, and hybrid schemes combining persistent and opportunistic access have not been investigated.

Open-source ns-3 support for NB-IoT evolved from early energy-aware LTE models [21] to LENA-NB [12], which implements Rel-13/14 features including cross-subframe scheduling. However, existing implementations support neither energy harvesting operation, supercapacitor storage, and brown-out behavior, nor persistent and FUG UL scheduling, preventing a system-level investigation of UL access schemes under realistic conditions.

System model

We consider a single-cell NB-IoT UL scenario in which a population of A-IoT UEs periodically reports short status messages to a remote application server. We assume that the

UEs are immobile and uniformly distributed around the BS within a cell of radius RR meters. The remote host is connected to the core network via a high-capacity point-to-point link.

The traffic generated by each UE is modeled as a two-state Markov-modulated source S={INACTIVE,ACTIVE}\mathcal{S} = \{\mathrm{INACTIVE}, \mathrm{ACTIVE}\}, with state transitions occurring every TT seconds. The transition probabilities are denoted by pIAp_{\mathrm{IA}} for the transition from INACTIVE to ACTIVE, and by pAIp_{\mathrm{AI}} for the reverse transition. Therefore, the steady-state probability of being in the ACTIVE state is πa=pIA/(pIA+pAI)\pi_a = p_{\mathrm{IA}}/(p_{\mathrm{IA}}+p_{\mathrm{AI}}). A UE generates a User Datagram Protocol (UDP) packet of size LL bytes at a decision epoch whenever it is in the ACTIVE state. Accordingly, the long-term mean offered traffic rate is 8πaL/T8\pi_a L/T bit/s per UE.

Additionally, each UE is equipped with an energy-harvesting module comprising a supercapacitor replenished by a time-varying solar energy source. The harvested input power follows a diurnal sin⁡2\sin^2 profile based on the photovoltaic power-system model of [22],

Ph(t)=Phmax⁡sin⁡2(π(t/Tday+ϕ)),(1)P_h(t) = P_h^{\max} \sin^2\left(\pi(t/T_{\mathrm{day}}+\phi)\right), \tag*{(1)}

where Phmax⁡P_h^{\max} denotes the peak harvested power, TdayT_{\mathrm{day}} the harvesting period, and ϕ\phi a configurable phase offset. The resulting profile has an average harvested power equal to half of Phmax⁡P_h^{\max}.

The supercapacitor, with capacitance CC, is modeled using the single-branch equivalent circuit of [23], comprising an ideal capacitor in series with an equivalent series resistance RsR_s and a parallel leakage resistance RleakR_{\mathrm{leak}} accounting for self-discharge. The radio is represented as a current sink rather than a resistive load, with iloadi_{\mathrm{load}} obtained from the measured power of the active radio state, so that consumption and self-discharge enter the model through separate terms. The capacitor state is updated in an event-driven manner at every radio power-state transition, harvest update, and at least every Δtpoll\Delta t_{\mathrm{poll}}, assuming a piecewise-constant net current over each update interval. Given the net current ik=iin(tk)−iload(tk)i_k = i_{\mathrm{in}}(t_k)-i_{\mathrm{load}}(t_k) over an interval of duration Δt\Delta t, the capacitor voltage evolves according to the closed-form solution of the first-order circuit,

v(tk+Δt)=Reqik(1−e−Δt/τ)+v(tk)e−Δt/τ,(2)v(t_k+\Delta t)=R_{\mathrm{eq}}i_k\left(1-e^{-\Delta t/\tau}\right)+v(t_k)e^{-\Delta t/\tau}, \tag*{(2)}

where τ=ReqC\tau=R_{\mathrm{eq}}C and ReqR_{\mathrm{eq}} is dominated by RleakR_{\mathrm{leak}} because Rleak≫RsR_{\mathrm{leak}}\gg R_s. Since every update interval satisfies Δt≪τ\Delta t\ll\tau, (2) reduces to Δv≈ikΔt/C−v(tk)Δt/τ\Delta v\approx i_k\Delta t/C-v(t_k)\Delta t/\tau, that is, charge accounting for the net current plus a slow self-discharge term. Energy drawn by the radio is therefore carried by iloadi_{\mathrm{load}} rather than by the time constant.

The voltage is clamped to [0,vmax⁡][0,v_{\max}], and the stored energy is E(t)=12Cv2(t)E(t)=\frac{1}{2}Cv^2(t). Device operation is governed by a voltage threshold vthv_{\mathrm{th}}. When v(t)<vthv(t)<v_{\mathrm{th}}, the UE enters a brown-out state in which radio transmission and protocol operations are suspended and no packets are generated. The connection context is retained, so recovery does not require a new attachment. Operation resumes once the voltage exceeds vrec=vth+Δvv_{\mathrm{rec}}=v_{\mathrm{th}}+\Delta v, where Δv\Delta v introduces hysteresis to avoid oscillatory switching near the threshold. This energy gating couples the harvesting process with both traffic generation and radio-access activity.

A packet is thus transmitted only when the UE is in the ACTIVE traffic state and not in brown-out, and an uplink grant has been obtained through the configured access scheme. The energy consumed is determined by the UE radio state, with each state in Table I assigned a constant power draw derived from practical NB-IoT measurements [24]. The system model, coupling stochastic traffic, harvesting dynamics, and radio access, is illustrated in Figure 2.

System model overview

Figure 2. System model overview: two-state Markov-modulated per-UE traffic source (top-left); per-UE energy front-end with ambient harvester Ph(t)P_h(t) feeding a leaky supercapacitor (C,v(t))(C,v(t)) under voltage-threshold brown-out gating (top-right); single-cell deployment of A-IoT UEs (dashed blue: UL) around the BS, connected to a remote application server through the core network via a fixed-delay backhaul (dashed gray).

ParameterSymbolValueParameterSymbolValue
Cell radiusRR1.25 kmDCI 0 scheduling delayk0k_08 ms
Number of UEsNUEN_{\mathrm{UE}}1–200Contention resolutionTCRT_{\mathrm{CR}}512 ms
UE / BS Tx powerPUEt,PBStP_{\mathrm{UE}}^{t}, P_{\mathrm{BS}}^{t}10, 30 dBmNPDCCH USS configRmax⁡USS,GUSSR_{\max}^{\mathrm{USS}}, G^{\mathrm{USS}}8, 2 (TUSS=16T_{\mathrm{USS}}=16 ms)
Noise figureNFNF0 dBPSM power drawPPSMP_{\mathrm{PSM}}15 μ\muW
Pathloss model–FriiseDRX / connected-idle power drawPidleP_{\mathrm{idle}}3 mW
P2P link delayτp2p\tau_{\mathrm{p2p}}10 msNPDCCH / NPDSCH Rx power drawPrxP_{\mathrm{rx}}80 mW
Application payloadLL49 BNPRACH / NPUSCH Tx power drawPtxP_{\mathrm{tx}}589 mW
State update intervalTT300 sCapacitanceCC1 F
Transition probabilitiespIA,pAIp_{IA},p_{AI}0.7, 0.2Initial / Max voltagev0,vmax⁡v_0,v_{\max}∼U(2.9,3.3)\sim U(2.9, 3.3), 3.3 V
RLC mode–AMCut-off / recovery voltagevth,vrecv_{\mathrm{th}},v_{\mathrm{rec}}2.4, 2.6 V
Early Data Transmission–enabledInternal / leakage RRRs,RleakR_s,R_{\mathrm{leak}}0.1, 10410^4 Ω\Omega
NPRACH periodicity (CE-0)TNPRACHT_{\mathrm{NPRACH}}80 msHarvest profilePh(t)P_h(t)sin⁡2\sin^2, eq. (1)
NPRACH / reserved SR subcarriersPsc,PscSRP_{\mathrm{sc}},P_{\mathrm{sc}}^{\mathrm{SR}}12, 6Peak harvest powerPhmax⁡P_h^{\max}1, 1.5, 2 mW
SR reservation validityTresT_{\mathrm{res}}3 hEnergy poll intervalΔtpoll\Delta t_{\mathrm{poll}}100 ms
NPRACH reps / max attemptsNrepN_{\mathrm{rep}}, –1, 10Active / TAU timerT3324,T3412T_{3324},T_{3412}2 s, 24 h
RAR windowWRARW_{\mathrm{RAR}}160 mseDRX cycle lengthTeDRXT_{\mathrm{eDRX}}40.96 s
MSG3 scheduling delayk2k_212 msData-inactivity releaseTrelT_{\mathrm{rel}}1 s
NPUSCH durationTNPUSCHT_{\mathrm{NPUSCH}}4 msSimulation durationTsim=TdayT_{\mathrm{sim}}=T_{\mathrm{day}}1 h

Table I. Simulation parameters; per-state power draw from Quectel BG96 measurements [24].

Extension of the ns-3 NB-IoT Module

We extend the LENA-NB ns-3 platform [12] with the components required to support active A-IoT operation together with considered UL access schemes. The resulting framework augments the NB-IoT protocol stack with energy-constrained traffic generation, supercapacitor-based energy storage, SR, HSR and FUG.

To model the intermittent traffic generated by A-IoT devices, we introduced a new application-layer class, MarkovUdpClient. This class implements the two-state Markov-modulated UDP source described in IV. Energy storage is modeled through a new GenericCapacitor class derived from EnergySource. The implementation captures RC supercapacitor dynamics, including leakage effects, internal resistance, voltage clamping at vmax⁡v_{\max}, and closed-form capacitor-state evolution. Harvested energy is supplied by the SolarEnergyHarvester class derived from EnergyHarvester, which injects the diurnal power profile Ph(t)P_h(t) defined in (1) into the capacitor. The NbiotEnergyModel class was extended with a hysteretic brown-out mechanism and per-radio-state energy accounting. It also exports a callback interface that notifies the MarkovUdpClient application of brown-out events.

To support persistent UL access and FUG, the RRC and Medium Access Control (MAC) layers were extended to retain the UE context across sleep intervals. The LteEnbRrc and

LteUeRrc classes implement configurable massive-IoT timers (i.e., T3324T_{3324}, T3412T_{3412}, TeDRXT_{\mathrm{eDRX}}), context retention, suspend/resume procedures and Release Assistance Indication (RAI)-assisted connection release. Suspension is synchronized with the Radio Link Control (RLC) acknowledgement state to ensure that pending acknowledgements are exchanged before the connection is suspended. LteUeMac implements dedicated SR using reserved NPRACH resources and opportunistic contention on shared resources for HSR. The corresponding BS-side resource mapping, contention resolution, and EDT procedures are implemented in LteEnbMac, LteEnbRrc, and NbiotScheduler. For proactive FUG, NbiotScheduler schedules an uplink grant for each UE at the predicted next transmission time. The added and modified classes are summarized in Figure 3. The extended module, evaluation scenario, and analysis scripts are publicly available on Github [25].

Abridged ns-3 NB-IoT models extended for ambient IoT

Figure 3. Abridged ns-3 NB-IoT models extended for ambient IoT.

Performance Evaluations

Simulation setup

UEs start pre-registered but suspended in an idle state, such that the first UL packet triggers the standard NB-IoT RA procedure and establishes the connection context. Connected UEs are released after a data-inactivity timeout of Trel=1T_{\mathrm{rel}}=1 s and transition through the A-IoT timer profile (i.e., T3324=2T_{3324}=2 s, TeDRX=40.96T_{\mathrm{eDRX}}=40.96 s, and T3412=24T_{3412}=24 h) into PSM. This life cycle is common to all schemes, including Idealized FUG. For the later, each packet arrival is known to the BS instantly through a Buffer Status Report (BSR) of zero latency and signaling cost. Since a packet arrival coincides with the UE wake-up, the BS schedules the UL grant at the next available NPDCCH opportunity. EDT is enabled throughout, allowing each contention-based access to carry the application payload

Protocol processing follows the default NB-IoT stack, with Packet Data Convergence Protocol (PDCP) and RLC operating in acknowledged mode and without header compression. NPRACH opportunities recur every 80 ms, while the scheduler allocates resources at each USS occasion (TUSS=16T_{\mathrm{USS}}=16 ms), using a round-robin policy. All UL transmissions employ single-tone 15 kHz NPUSCH Format 1. Under SR and HSR, six of the twelve Coverage Enhancement (CE)-0 NPRACH subcarriers are reserved for dedicated SR preambles (PscSR=6P_{\mathrm{sc}}^{\mathrm{SR}}=6) while the remaining six are available for contention-based access and the initial attachment procedure. Propagation is modeled using the Friis path-loss model. Since the simulated cell dimension (R=1.25R=1.25 km) is relatively small for NB-IoT networks under free space propagation, all UEs operate under the same maximal Modulation and Coding Scheme (MCS) and the same CE-0. This isolates the comparison from link-adaptation and coverage-extension effects that would otherwise depend on UE placement.

Each UE is powered by a 1 F supercapacitor (vmax⁡=3.3v_{\max}=3.3 V, vth=2.4v_{\mathrm{th}}=2.4 V) replenished by the diurnal harvesting profile of (1) with a peak harvested power varied across 1, 1.5 and 2 mW. The initial voltage is drawn independently per device from v0∼U(2.9,3.3)v_0\sim U(2.9,3.3) V to avoid synchronized cold-start. A hysteretic brown-out mechanism with a recovery threshold of vrec=2.6v_{\mathrm{rec}}=2.6 V disables both radio activity and application traffic until the capacitor voltage exceeds the recovery threshold. The per-state radio power consumption follows the Quectel BG96 measurements. The complete set of simulation parameters is summarized in Table I.

Discussion of results

We evaluate the access schemes using network-level and energy-related metrics. Each data point averages ten independent runs with different seeds; error bars indicate standard deviation. Reliability is measured by the mean UL delay Dˉ\bar{D} over delivered packets. Energy efficiency is quantified by the energy per delivered bit ϵ=Ealive/Brx\epsilon=E_{\mathrm{alive}}/B_{\mathrm{rx}}, where EaliveE_{\mathrm{alive}} is the energy consumed by a UE while it is not brown-out and BrxB_{\mathrm{rx}} the delivered payload. Sustainability is evaluated by the brown-out fraction ϕ\phi, the fraction of devices whose capacitor voltage falls below vthv_{\mathrm{th}} at least once.

Network-level metrics

  1. Network-level metrics: Figure 4a shows the corresponding mean UL delay Dˉ\bar{D}. The Idealized FUG provides a lower bound, with delay increasing from 50 ms to 212 ms as grant deliveries serialize over shared NPDCCH occasions at high density. Among the deployable schemes, HSR exhibits the slowest increase in delay as the number of UEs grows, from 222 ms at NUE=20N_{\mathrm{UE}}=20 to 595 ms at NUE=200N_{\mathrm{UE}}=200. This limited

Performance versus number of UEs, showing average uplink delay, energy per bit, and fraction of depleted UEs

Figure 4. Performance vs. the number of UEs (Phmax⁡=2P_h^{\max}=2 mW for (a), (b)): (a) mean uplink delay Dˉ\bar{D}, (b) energy per delivered bit ϵ\epsilon. (c) Fraction of UEs with at least one brown-out in Msg3. Traffic follows the Markov source of Sec. IV with T=300T=300 s, (pIA,pAI)=(0.7,0.2)(p_{IA},p_{AI})=(0.7,0.2), L=49L=49 B.

increase results from the opportunistic contention path, which resolves most requests before the dedicated request occasion is reached. In contrast, the mean delay of SR increases from 222 ms to 1.43 s as the round-robin request period increases with the number of UEs. The delay of RA similarly increases from 177 ms to 1.60 s as contention intensifies. At NUE=200N_{\mathrm{UE}} = 200, HSR therefore reduces the mean UL delay by a factor of 2.4–2.7 relative to SR and RA, respectively.

Energy efficiency and sustainability

  1. Energy efficiency and sustainability. Figure 4b shows the energy per delivered bit ε\varepsilon. Beyond NUE≈180N_{\mathrm{UE}} \approx180, idealized FUG becomes less energy-efficient than SR, as increasing grant serialization extends NPDCCH monitoring for waking UEs. SR is the most energy-efficient among the deployable schemes and remains nearly insensitive to the number of UEs, with ε\varepsilon ranging from 0.234 mJ/bit to 0.238 mJ/bit. Dedicated, collision-free request opportunities keep the per-packet awake time nearly constant, while the waiting time for the reserved occasion is spent in PSM. RA consumes 17–31 % more energy per delivered bit than SR, increasing from 0.27 mJ/bit to 0.31 mJ/bit as contention increases the time spent accessing the channel. HSR has energy consumption close to SR at low density but increases to 0.33 mJ/bit at NUE=200N_{\mathrm{UE}} = 200, exceeding RA at approximately NUE=170N_{\mathrm{UE}} = 170. At high density, three effects contribute to this increase: nearly every packet contends before its reserved request occasion, the contention request partition is half the size of the RA pool, and each failed attempt incurs a full response-window monitoring period. Thus, HSR’s lower request latency comes at the cost of increased contention energy, whereas SR maintains nearly constant energy consumption by trading the same increase in density for request delay.

Figure 4c reports the brown-out fraction ϕ\phi. At a peak harvested power of 2 mW, all schemes complete every run without brown-out, with ϕ≈0\phi\approx0. At 1.5 mW, differences in energy consumption become apparent: SR maintains a nearly load-independent ϕ≈0.08\phi\approx0.08, whereas RA increases from 0.23 to 0.35 and HSR from 0.12 to 0.37. The latter exceeds RA at high density, consistent with its higher contention energy. At 1 mW, none of the schemes can sustain operation, with ϕ\phi ranging from 0.60 to 0.95. Thus, brown-out becomes significant as the harvested power decreases below the level required for sustainable operation.

Conclusion

We studied the energy–latency trade-off of UL access for energy-harvesting A-IoT devices by comparing RA, SR, the proposed HSR, and FUG. The Idealized FUG configuration establishes a lower bound on achievable delay and energy cost when future traffic is known perfectly, while practical FUG performance degrades with traffic prediction errors. Among the implementable access schemes, SR provides the best energy efficiency, achieving approximately 0.24 mJ/bit, 17 %–31 % below RA. In contrast, HSR provides the lowest delay, reducing mean UL delay by a factor of 2.4–2.7×\times relative to SR and RA at NUE=200N_{\mathrm{UE}} = 200, respectively. This delay reduction comes at the cost of additional contention energy, causing HSR to become less energy-efficient than RA beyond approximately 170 UEs. The preferred UL access scheme therefore depends on network density, latency requirements, and available harvested energy. Future work will develop heuristic and learning-based predictive FUG schedulers that operate without perfect traffic knowledge and evaluate how closely they can approach the delay and energy lower bounds established by Idealized FUG.

Acknowledgement

This work was funded by the AMBIENT-6G project of the Smart Networks and Services Joint Undertaking (SNS JU), EU Horizon Europe Grant No. 101192113.

References

  1. [1]3GPP, “Study on A-IoT in RAN,” 3GPP, TR 38.848, 2023, V18.0.0.
  2. [2]3GPP, “NR; A-IoT physical layer,” 3GPP, TS 38.291, 2026, V19.3.0.
  3. [3]3GPP, “NR; A-IoT MAC protocol,” 3GPP, TS 38.391, 2026, V19.4.0.
  4. [4]3GPP, “Solutions for ambient IoT (Internet of Things) in NR Phase 2,” 3GPP, Work Item Description RP-260826, 2026.
  5. [5]M. M. Butt et al., “Ambient IoT: A Missing Link in 3GPP IoT Devices Landscape,” IEEE IoT Magazine, DOI: 10.1109/IOTM.001.2300198
  6. [6]F. Michelinakis et al., “Dissecting Energy Consumption of NB-IoT Devices Empirically,” IEEE IoT J, DOI: 10.1109/JIOT.2020.3013949DOI
  7. [7]A. Höglund et al., “3GPP Rel-16 PUR for LTE-M and NB-IoT,” IEEE Comms. Stds. Mgzine, 2020. DOI: 10.1109/MCOMSTD.001.2000003
  8. [8]G. A. Medina-Acosta et al., “3GPP Rel-17 PHY Enhancements for LTE-M and NB-IoT,” IEEE Comms Stds Mgzine, vol. 6, pp. 80–86, 2022. DOI: 10.1109/MCOMSTD.001.2200055
  9. [9]S. Ali et al., “FUG for MTC.: Challenges and Opportunities,” IEEE Comm. Mgzne, DOI: 10.1109/MCOM.2018.1800813
  10. [10]3GPP RAN1, “Draft minutes report, RAN1 meeting #126,” 3GPP, Tech. Rep., 2026, Agenda item 9.3.
  11. [11]3GPP RAN2, “Draft meeting report, RAN2 meeting #135,” 3GPP, Tech. Rep., 2026, Agenda item 8.2.
  12. [12]P. Jörke et al., “From LENA to LENA-NB: Impl. and Perf. Eval. of NB-IoT and EDT in ns-3,” in Proc. Wrkshp on ns-3, 2022. DOI: 10.1145/3532577.3532600DOI
  13. [13]A. K. Sultania et al., “Energy Modeling and Eval. of NB-IoT with PSM and eDRX,” in Proc. IEEE GLBECM Wrksphs, 2018. DOI: 10.1109/GLOCOMW.2018.8644074DOI
  14. [14]J. P. García-Martín et al., “Energy consumption analytical modeling of NB-IoT devices for diverse IoT applications,” Computer Networks, DOI: https://doi.org/10.1016/j.comnet.2023.109855
  15. [15]Y. Liu et al., “Analysis of RA in NB-IoT Networks With Three Coverage Enhancement Groups: A Stochastic Geometry Approach,” IEEE Trans. on Wireless Comm., DOI: 10.1109/TWC.2020.3026331DOI
  16. [16]3GPP TS 36.213: E-UTRA; Physical layer procedures, 3GPP, 2020.
  17. [17]S. Ali et al., “Sleeping MAB for FUG Allocation in MTC,” in 2018 IEEE GC Wkshps. DOI: 10.1109/GLOCOMW.2018.8644350
  18. [18]E. Eldeeb et al., “Traffic prediction and fast uplink for hidden markov IoT models,” arXiv preprint arXiv:2209.05939, 2022.DOI
  19. [19]T.-Y. Wu et al., “Persistent Periodic Uplink Scheduling Algorithm for Massive NB-IoT Devices,” Sensors, vol. 22, no. 8, 2022, ISSN: 1424-8220. DOI: 10.3390/s22082875DOI
  20. [20]L. Moons et al., “Evaluating Fast and Grant-Free Uplink Access in Next-Generation Cellular IoT Networks,” in 2024 3rd 6GNet. DOI: 10.1109/6GNet63182.2024.10765754DOI
  21. [21]A. K. Sultania et al., “Impl. of NB-IoT Power Saving Schemes in ns-3,” in Proc. Wrkshp on ns-3, 2019. DOI: 10.1145/3337941.3337944DOI
  22. [22]D. Rekioua et al., “Optimization of Photovoltaic Power Systems: Modelization, Simulation and Control. Jan. 2012, vol. 102, ISBN: 978-1-4471-2348-4. DOI: 10.1007/978-1-4471-2403-0DOI
  23. [23]R. L. Spyker et al., “Classical equivalent circuit parameters for a double-layer capacitor,” IEEE Trans. on Aerospace and Electronic Systems, vol. 36, no. 3, pp. 829–836, Jul. 2000. DOI: 10.1109/7.869502DOI
  24. [24]Quectel Wireless Solutions, BG96 LTE Cat M1 / Cat NB1 / EGPRS Module Hardware Design Document, document version V1.6, Quectel, 2020.
  25. [25]D. D. Agbeve, ns3 nbiot for aiot. [Online]. Available: https://github.com/imec-idlab/ns3-nbiot-ambient-aiot

Paper details

vlnc:W_3xpqj53r
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Not recorded for this paper

View statement
Looking for related papers…
Contents