---
title: Multi-Beam Time-Frequency Slot Assignment
url: https://www.emergentmind.com/topics/multi-beam-time-frequency-slot-assignment-mb-tfsa
type: topic
---

# Multi-Beam Time-Frequency Slot Assignment

Searching arXiv for the cited MB-TFSA-related papers to ground the article in current records.
Multi-Beam Time-Frequency Slot Assignment (MB-TFSA) denotes the assignment of communication opportunities across multiple beams, time slots, and, where applicable, frequency channels or other time-frequency structures, under coupling constraints induced by interference, beam exclusivity, power budgets, quality-of-service targets, and physical-layer beamforming or detection. Across the cited literature, the term covers several closely related formulations: downlink OFDMA-SDMA scheduling with zero-forcing precoding, GEO high-throughput satellite user scheduling with full frequency reuse, return-link MF-TDMA scheduling with free-slot assignment, dynamic LEO beam-direction and subchannel-time allocation, time-frequency pilot scheduling in massive MIMO-OFDM, and QUBO-based flow scheduling for digital satellites [1212.5284].

## 1. Core definition and problem abstractions

In its canonical form, MB-TFSA assigns users, flows, or pilot patterns to resource units indexed by beam, time, and frequency. In the digital-satellite flow-scheduling formulation, a resource unit is a tuple $u=(b,f,s)\in U=B\times F\times S$, where $B$ is the set of beams, $F$ the set of frequency channels, and $S$ the set of time slots; the binary decision variable $x_{k,u}\in\{0,1\}$ indicates whether flow $k$ is served on resource unit $u$ [2603.00701]. In dynamic LEO systems, the corresponding time-frequency items are subchannel-time-slot units $(k,t)\in B_{SA}=K\times T$, assigned through $b^t_{k,n}\in\{0,1\}$, while beam pointing is represented separately by coordinate-time-slot units $(c,t)\in C\times T$ via $e^t_{c,q}\in\{0,1\}$ [2401.09711]. In OFDMA-SDMA downlink systems, the assignment variable is $x_{k,n,t}\in\{0,1\}$, indicating whether user $k$ is scheduled on subcarrier $n$ at time $t$, with beam occupancy implicit in the precoder columns associated with the scheduled set $S_{n,t}$ [1212.5284].

Several works instantiate only a restricted version of MB-TFSA. In the GEO precoded MB-HTS setting with full frequency reuse, there is effectively a single frequency slot, so MB-TFSA reduces to multi-beam time-slot assignment, or MB-TSA, over an observed window time of $T$ time slots [2110.02525]. In the multibeam frequency-management framework for mobile users, the baseline problem is frequency assignment over a planning horizon $[0,T]$, and the MB-TFSA interpretation arises by discretizing the time axis into slots $t\in T$ and allocating discrete frequency channels $f\in F$ per slot [2403.07007]. In the massive MIMO-OFDM pilot-design setting, MB-TFSA corresponds to assigning time-frequency phase shifts $\{\phi_u,\varphi_u\}$ and available TF slots or pilot segments so that equivalent triple-beam supports do not overlap [2505.04933].

The main structural commonality is that the resource-assignment decision is never isolated. It is always coupled to another layer: beamforming and power loading in OFDMA-SDMA [1212.5284], linear precoding and QoS enforcement in GEO forward-link MB-HTS [2110.02525], MMSE-SIC detection and free-slot assignment in the return link [1210.4329], beam direction control and power allocation in LEO [2401.09711], pilot-phase design and tensor estimation in massive MIMO-OFDM [2505.04933], or slack-bit-augmented QUBO penalties in digital satellites [2603.00701]. A plausible implication is that MB-TFSA is best viewed as a family of coupled combinatorial resource-allocation problems rather than a single fixed formulation.

## 2. Physical-layer models and resource semantics

In downlink OFDMA-SDMA with zero-forcing precoding, a single base station equipped with $M$ transmit antennas serves $K$ single-antenna users over $N$ OFDMA subcarriers across time slots $t=1,2,\ldots$, and up to $M$ users can be multiplexed simultaneously in each slot $(n,t)$ by linear precoding with zero-forcing constraints [1212.5284]. For a scheduled set $S_{n,t}\subseteq\{1,\ldots,K\}$ with $|S_{n,t}|\le M$, the received signal is
$$
y_{k,n,t}=h_{k,n,t}^H w_{n,t,k}\,x_{n,t,k}
+\sum_{j\in S_{n,t},\,j\neq k} h_{k,n,t}^H w_{n,t,j}\,x_{n,t,j}+z_{k,n,t},
$$
with $h_{k,n,t}^H w_{n,t,j}=0$ for all $k\neq j\in S_{n,t}$ under ZF, yielding
$$
\mathrm{SINR}_{k,n,t}=\frac{|h_{k,n,t}^H w_{n,t,k}|^2}{\sigma^2},\qquad
R_{k,n,t}=\log_2\!\big(1+\mathrm{SINR}_{k,n,t}\big).
$$
Once $S_{n,t}$ is fixed, beam design collapses to power loading through effective gains
$$
g_{k,n,t}=\frac{1}{\big[(H_{n,t}^H H_{n,t})^{-1}\big]_{kk}},
$$
with $H_{n,t}=[h_{k,n,t}]_{k\in S_{n,t}}$ [1212.5284].

In GEO multi-beam high-throughput satellite systems with full frequency reuse, a single gateway precodes the signals to all beams that reuse the same band. The per-slot scheduled channel matrix is $H(t)=[h_{\pi_1},\ldots,h_{\pi_{|K(t)|}}]\in\mathbb{C}^{M\times |K(t)|}$, and the paper uses regularized zero-forcing precoding,
$$
W(t)=\frac{1}{\sqrt{\omega(t)}}\,H(t)\big(H(t)^H H(t)+\alpha(t)I\big)^{-1},
$$
with $\alpha(t)=\sigma^2 |K(t)| / P_{\max}$ [2110.02525]. With symbols $s_k(t)$, powers $p_k(t)\ge 0$, and noise $n_k(t)\sim \mathcal{CN}(0,\sigma^2)$, the received signal is
$$
y_k(t)=\sum_{k'\in K(t)} \sqrt{p_{k'}(t)}\, h_k^H w_{k'}(t)\, s_{k'}(t)+n_k(t),
$$
and
$$
\mathrm{SINR}_k(t)=
\frac{p_k(t)|h_k^H w_k(t)|^2}{\sum_{k'\in K(t)\setminus\{k\}} p_{k'}(t)|h_k^H w_{k'}(t)|^2+\sigma^2},
\qquad
R_k(t)=B\log_2(1+\mathrm{SINR}_k(t)).
$$
Because reuse factor is $1$, interference coupling is across all beams that are simultaneously active in the same slot [2110.02525].

In the multibeam return link with universal frequency reuse, MB-TFSA is framed as gateway-controlled scheduling of return-link transmissions across multiple spot beams into MF-TDMA slots, with one user per active beam per slot and optional free slots to lower instantaneous co-channel interference [1210.4329]. The slot-level received model is
$$
\mathbf{y}_s=\mathbf{H}_s\,\mathbf{x}_s+\boldsymbol{\eta}_s,
$$
and multiuser detection is performed by MMSE filtering with optimal successive interference cancellation ordering [1210.4329].

In dynamic LEO networks, the physical model additionally includes beam-direction control. The transmit antenna gain from beam $q$ pointing to center $c$ toward user $n$ on subchannel $k$ at time $t$ is
$$
G^{tx,t}_{q,c,k,n}=G_q^{peak}\left(\frac{J_1(\mu)}{2\mu}+\frac{36J_3(\mu)}{\mu^3}\right)^2,
$$
with $G_q^{peak}=\eta(\pi D f / c)^2$ and $\mu=2.07123 \sin(\theta^t_{q,c,n})/\sin(\theta^t_{q,3dB})$ [2401.09711]. The achievable rate on $(q,c,k,n,t)$ is
$$
R^t_{q,c,k,n}=B_{q,k}\log_2(1+\gamma^t_{q,c,k,n}),
$$
where $B_{q,k}=B_q/K$ and $\gamma^t_{q,c,k,n}$ includes inter-beam interference from simultaneous reuse on the same subchannel [2401.09711].

A common source of ambiguity is that “beam” need not always mean a transmit beam serving payload traffic. In the triple-beam pilot-scheduling model, “triple-beam” refers to simultaneous structure across spatial, frequency, and time beam matrices $V^s$, $V^f$, and $V_T^t$, with the space-frequency-time channel tensor expressed as
$$
\mathcal{H}_{u,T}^{SFT}=\mathcal{V}_T *_{3} \mathcal{H}_u^{TB},\qquad
\mathcal{H}_{u,T}^{SFT,p}=\mathcal{V}_T^p *_{3} \mathcal{H}_u^{TB},
$$
for user $u$ [2505.04933]. This suggests that, in some contexts, MB-TFSA generalizes from data scheduling to structured pilot-domain assignment.

## 3. Optimization formulations and constraints

The optimization problems associated with MB-TFSA are consistently mixed-integer and non-convex. In OFDMA-SDMA with ZF, the per-slot resource-allocation problem is
$$
\begin{aligned}
\max_{\{x_{k,n,t},\,w_{n,t,k}\}} \quad &
\sum_{k=1}^{K}\sum_{n=1}^{N} x_{k,n,t}\,
\log_2\!\Big(1+\tfrac{|h_{k,n,t}^H w_{n,t,k}|^2}{\sigma^2}\Big) \\
\text{s.t.}\quad &
\sum_{n=1}^{N}\sum_{k=1}^{K}\|w_{n,t,k}\|_2^2 \le P_{\mathrm{tot},t},\\
&
\sum_{n=1}^{N} x_{k,n,t}\,
\log_2\!\Big(1+\tfrac{|h_{k,n,t}^H w_{n,t,k}|^2}{\sigma^2}\Big)\ge R_k^{\min}
\quad \forall k\in\mathcal{K}_{\mathrm{RT}},\\
&
h_{k,n,t}^H w_{n,t,j}=0 \quad \forall n,\ \forall k\neq j\in S_{n,t},\\
&
\sum_{k=1}^{K} x_{k,n,t}\le M \quad \forall n,\qquad
x_{k,n,t}\in\{0,1\},
\end{aligned}
$$
and is described as nonlinear, non-convex, and mixed-integer due to ZF structure and binary assignments [1212.5284].

In GEO MB-HTS with individual QoS constraints, scheduling and power allocation are coupled through per-user aggregate data demands $\xi_k$ over a window of $T$ slots. The paper enforces a per-slot surrogate QoS condition
$$
R_k(t)\ge \xi_k/T_k
$$
for each slot in which user $k$ is scheduled, guaranteeing $R_{k,\mathrm{agg}}\ge \xi_k$ if the user occupies $T_k$ slots [2110.02525]. For fixed scheduled sets $K(t)$, the power-allocation subproblem remains nonconvex because SINR constraints and objective terms involve sums of ratios of posynomials [2110.02525].

In the return-link MF-TDMA case, the core MB-TFSA formulation uses binary assignment variables $x_{u,b,s}\in\{0,1\}$ and optional free-slot indicators $f_{b,s}\in\{0,1\}$, with objective
$$
\max_{\{x,f\}}\ \min_u R_u,
$$
subject to at most one user or a free slot per beam and slot,
$$
\sum_u x_{u,b,s}\le 1-f_{b,s},
$$
at most one slot per scheduled user,
$$
\sum_s x_{u,b,s}\le 1,
$$
and, under free-slot assignment, one free slot per beam across $M+1$ slots,
$$
\sum_s f_{b,s}=1.
$$
The rates depend jointly on the selected multi-beam path and SIC order, so the problem is combinatorial and non-separable across beams [1210.4329].

In LEO systems, the master problem jointly optimizes subchannel-time allocation $b^t_{k,n}$, beam-direction variables $e^t_{c,q}$, and powers $p^t_{m,q}$ to maximize the long-term fairness-aware objective
$$
\sum_{n\in N} U^\alpha\!\left(\sum_{t\in T}\sum_{q\in Q} R^t_{q,n}\right),
$$
where
$$
U^\alpha(x)=
\begin{cases}
(1-\alpha)^{-1}x^{1-\alpha}, & 0\le \alpha<1,\\[4pt]
\log(x), & \alpha=1.
\end{cases}
$$
The constraints include within-beam OFDMA orthogonality, per-user subchannel caps, one center per beam and slot, one beam per center and slot, minimum elevation angle, minimum SINR, per-beam power limits, and per-satellite power limits [2401.09711].

In digital satellites, the pre-QUBO MB-TFSA formulation is a binary weighted-throughput maximization,
$$
\max \sum_{k\in K}\sum_{u\in U} w_k r_{k,u} x_{k,u},
$$
subject to resource exclusivity,
$$
\sum_{k\in K} x_{k,u}\le 1,\quad \forall u\in U,
$$
per-slot power budgets,
$$
\sum_{u\in U_s}\sum_{k\in K} p_u x_{k,u}\le P_{\max}(s),\quad \forall s\in S,
$$
and per-flow queue-capacity constraints,
$$
\sum_{u\in U} r_{k,u} x_{k,u}\le C_k,\quad \forall k\in K
$$
[2603.00701]. Optional interference and guard-band extensions are noted but not instantiated in the core QUBO experiments [2603.00701].

The following variants summarize the range of formulations.

| Setting | Assignment unit | Dominant constraints |
|---|---|---|
| OFDMA-SDMA with ZF | $(n,t)$ with scheduled set $S_{n,t}$ | ZF orthogonality, RT minimum rate, power budget, $|S_{n,t}|\le M$ |
| GEO MB-HTS forward link | Time slot $t$ | One user per beam per slot, QoS surrogate $R_k(t)\ge \xi_k/T_k$, sum power |
| Return-link MF-TDMA | Beam-slot pair $(b,s)$ | Max-min fairness, one user or free slot per beam per slot |
| Dynamic LEO | Subchannel-time $(k,t)$ plus coordinate-time $(c,t)$ | Interference-aware matching, beam direction, SINR, power |
| Digital satellite QUBO | Beam-frequency-time unit $(b,f,s)$ | Exclusivity, per-slot power, queue capacity |

## 4. Algorithmic approaches

The dual-based method for OFDMA-SDMA with ZF begins by relaxing the total power budget and real-time minimum-rate constraints with dual variables $\lambda\ge 0$ and $\mu_k\ge 0$ [1212.5284]. Using effective gains, the Lagrangian decomposes additively over time-frequency slots $(n,t)$, and, for a fixed candidate set $S_{n,t}$, the power-loading subproblem is
$$
\max_{\{p_{k,n,t}\ge 0\}_{k\in S_{n,t}}}
\sum_{k\in S_{n,t}} c'_k \log_2\!\Big(1+\frac{g_{k,n,t}p_{k,n,t}}{\sigma^2}\Big)
-\lambda \sum_{k\in S_{n,t}}\big[(H_{n,t}^H H_{n,t})^{-1}\big]_{kk}p_{k,n,t},
$$
with $c'_k=1+\mu_k$ for unit weights [1212.5284]. The optimal power is waterfilling-like,
$$
p_{k,n,t}^{\star}
=
\left[
\frac{c'_k}{\lambda \ln 2}\cdot \frac{\sigma^2}{\gamma_{k,n,t}^2}
-\frac{\sigma^2}{g_{k,n,t}}
\right]^+,
$$
and the scheduled set is chosen by maximizing the dual metric $\Phi(S_{n,t};\lambda,\mu)$ over all candidate sets with $|S_{n,t}|\le M$ [1212.5284]. Subgradient updates on $\lambda$ and $\mu_k$ produce a dual upper bound, while a primal-feasible recovery gives a lower bound [1212.5284].

In GEO MB-HTS systems, the paper proposes a successive optimization approach with two stages: heuristic scheduling with fixed powers and GP-based power allocation for the chosen scheduled sets [2110.02525]. The scheduling stage has Strict and Relax variants. It sorts users by channel norm, seeds each slot with the strongest user, tentatively adds candidates by recomputing the RZF precoder, and accepts additions subject to monotonic sum-rate growth and, in Strict mode, immediate per-slot QoS feasibility [2110.02525]. The power-allocation stage reformulates
$$
\max \prod_{k\in K(t)} (1+\mathrm{SINR}_k(\{p_k\}))
$$
using auxiliary variables $\gamma_k$, lower-bounds the signomial term by an AM-GM monomial, and solves the resulting geometric program iteratively, updating the AM-GM weights until convergence to a KKT point [2110.02525].

The return-link scheduling algorithm uses a bipartite-graph approach. A path is a $B$-tuple selecting one user from each beam for a slot, and for $M$ users per beam the number of candidate paths is
$$
N_{\text{paths}}=M^B.
$$
Each path is weighted by the minimum per-user rate under MMSE-SIC with optimal ordering, and a minimum-deletion algorithm selects $M$ feasible paths by iteratively deleting the worst path if feasibility is preserved, otherwise marking it as selected [1210.4329]. When a target minimum rate cannot be met, the algorithm enables free-slot assignment by spreading $M$ users over $M+1$ slots, with one free slot per beam [1210.4329].

In dynamic LEO networks, the problem is decomposed into beam direction control and time-slot allocation, user subchannel assignment, and beam power allocation [2401.09711]. The first two are handled by matching with externalities. Beam-direction control uses an initial deferred-acceptance phase followed by swap matching with externalities until an exchange-stable matching is reached [2401.09711]. Subchannel assignment applies per-beam deferred acceptance and then a cross-beam negotiation step that drops a reused subchannel from the weaker side when doing so increases total utility across beams at that slot [2401.09711]. Power allocation is then solved by successive convex approximation using the lower bound
$$
\log_2(1+\gamma)\ge \mu \log_2(\gamma)+\nu,
$$
with $\mu=\tilde{\gamma}/(1+\tilde{\gamma})$ and
$\nu=\log_2(1+\tilde{\gamma})-\mu\log_2(\tilde{\gamma})$ at the current iterate [2401.09711].

In the pilot-scheduling setting, the MB-TFSA algorithm is explicitly two-stage: DSatur-based grouping of users according to a graph whose edge weights depend on the overlap metric
$$
\eta(\mathcal{A},\mathcal{B})=
\frac{\sum_{i,j,k} [\mathcal{A}\odot \mathcal{B}]_{i,j,k}}
{\sqrt{\sum [\mathcal{A}]_{i,j,k}^2}\sqrt{\sum [\mathcal{B}]_{i,j,k}^2}},
$$
followed by greedy assignment of TF phase-shift pairs $\{\phi,\varphi\}$ to groups so that equivalent TB-domain supports remain approximately disjoint [2505.04933].

In the digital-satellite QUBO framework, the constrained problem is embedded in
$$
H_{\text{QUBO}}=H_{\text{obj}}+H_{\text{rc}}+H_{\text{pow}}+H_{\text{queue}},
$$
with objective term
$$
H_{\text{obj}}=-\lambda_{\text{obj}}\sum_{k\in K}\sum_{u\in U} w_k r_{k,u} x_{k,u},
$$
resource-conflict penalty
$$
H_{\text{rc}}=\lambda_{\text{rc}}\sum_{u\in U}\Big(\sum_{k\in K}x_{k,u}-1\Big)^2,
$$
and slack-bit-based penalties for power and queue inequalities [2603.00701]. The resulting Ising Hamiltonian is solved by QAOA with layer-wise training, optional warm starts, and SPSA-like stochastic hill climbing [2603.00701].

## 5. Performance characterizations and benchmarking results

The dual-bound framework in OFDMA-SDMA is explicitly presented as a benchmark for heuristic MB-TFSA policies [1212.5284]. In scenarios with $M=3$, $K=16$, $N=16$, and $P=20$ dBm, the dual-feasible lower bound was within $0.2$–$0.25\%$ of the upper bound for a single RT user with increasing minimum rate $80$–$120$ bps/Hz, while the weight-adjustment method had a gap of $3$–$10\%$ [1212.5284]. With increasing large-scale attenuation $(0,5,10)$ dB, the dual-feasible gaps were approximately $0.16\%$, $0.70\%$, and $0.82\%$, whereas the weight-adjustment gaps were approximately $9.5\%$, $33\%$, and $52\%$ [1212.5284]. As the number of RT users increased from $1$ to $7$, the dual-feasible gap grew from approximately $0.16\%$ to approximately $3.52\%$, while the weight-adjustment gap rose from approximately $3.5\%$ to approximately $22.7\%$ and failed to find feasible points for $6$–$7$ RT users [1212.5284].

In GEO MB-HTS, the numerical evaluation uses $M=7$ beams, $110$ users per beam, $T=500$ slots, bandwidth $B=500$ MHz, Ka-band carrier $19.95$ GHz, and sum power $P_{\max}\approx 18.45$ dBW [2110.02525]. Scheduling-only results show that Relax mode achieves the highest sum throughput, while Strict mode guarantees QoS for all scheduled users and yields the largest per-user throughput [2110.02525]. Adding GP-based power allocation produces gains of $20$–$47\%$ over fixed power, including example average gains of random $(+47\%)$, SUS $(+39\%$ sum, $+44\%$ per-user), proposed Strict $(+20\%)$, and proposed Relax $(+31\%)$ [2110.02525]. With power control, Algorithm 2 Relax reaches the highest average sum throughput, approximately $7300$ Mbps, while Algorithm 2 Strict yields the highest per-user rate, approximately $1403$ Mbps [2110.02525].

The return-link multibeam scheduler quantifies the effect of free-slot assignment on weakest-user performance and efficiency [1210.4329]. At SNR $15$ dB and $B=7$, without FSA the slot availability for meeting the target minimum rate is approximately $89$–$91\%$ at $100\%$ efficiency [1210.4329]. With FSA, the reported examples are: for $M_{\mathrm{FSA}}=2$ and $r_{\mathrm{th}}=3$ bits/s/Hz, slot availability $99.46\%$, efficiency approximately $93.84\%$, and FSA use approximately $6.56\%$; for $M_{\mathrm{FSA}}=3$ and $r_{\mathrm{th}}=3.3$, availability $98.89\%$, efficiency approximately $94.94\%$, and FSA use approximately $5.33\%$; and for $M_{\mathrm{FSA}}=4$ and $r_{\mathrm{th}}=3.41$, availability $99.20\%$, efficiency approximately $96.36\%$, and FSA use approximately $3.78\%$ [1210.4329]. The paper also reports that $M_{\mathrm{FSA}}=4$ nearly matches the min-rate CDF of scheduling without FSA at $M=7$, while delivering substantial complexity reduction [1210.4329].

In dynamic LEO networks, the integrated framework is evaluated with $16$ orbits, $30$ satellites per orbit, altitude $780$ km, inclination $45^\circ$, $N=50$ users, $T=100$, $K=20$ subchannels per beam, $K^{thr}=6$, $P_q^{\max}=200$ W per beam, and $P_m^{\max}=1200$ W per satellite [2401.09711]. The outer loop converges in a few iterations, and with $L=7$ beams per satellite it converges in under $5$ iterations [2401.09711]. Relative to baselines, the proposal improves the number of served users by up to two times and the sum user data rate by up to $68\%$, with beam direction optimization contributing larger gains than power allocation alone [2401.09711].

The dynamic frequency-assignment framework for mobile users reports that, in scenarios with more than $900$ beams, the method is able to serve over $99.97\%$ of the fixed and mobile users [2403.07007]. In the $330$-user high-uncertainty case, reserving $20\%$ spectrum yields $N_{\mathrm{served}}/N_{\mathrm{users}}\approx 0.9997$, while a configuration combining $10\%$ reserve with proactive protection achieves approximately $0.9994$ with approximately $41.5\%$ fewer reconfigurations [2403.07007].

In the pilot-scheduling problem, TFPSP-IGA achieves approximately $7$ dB NMSE reduction relative to APSP-IGA for $U=48$ at SNR $20$ dB, and more than $8$ dB reduction for $U=300$ at SNR $20$ dB [2505.04933]. The IGA method reaches target NMSE in approximately $180$ iterations at $20$ dB, whereas GAMP/EPV require more than $450$ iterations [2505.04933]. In the quantum MB-TFSA study, shallow QAOA with $p=1$ finds the optimal throughput for the $2$-flow and $4$-flow instances, while $p=3$ underperforms on hardware because increased depth raises noise and exacerbates barren plateaus [2603.00701].

## 6. Design implications, ambiguities, and limitations

A recurring design principle is decomposition. In OFDMA-SDMA, per-$(n,t)$ decomposition emerges after dualization and ZF collapse to effective gains [1212.5284]. In GEO MB-HTS, slotwise scheduling and per-slot GP-based power control are separated within a successive optimization loop [2110.02525]. In LEO, the decomposition into beam direction control, subchannel assignment, and power allocation is explicit [2401.09711]. In pilot scheduling, grouping and phase assignment are separated [2505.04933]. This suggests that exact end-to-end MB-TFSA is generally avoided in favor of structured decompositions that preserve tractability.

A second recurring issue is that interference is handled differently across domains. In downlink OFDMA-SDMA, zero-forcing removes intra-slot inter-user interference inside a scheduled set, but poor conditioning of $H_{n,t}$ inflates the ZF column norms and makes power loading expensive [1212.5284]. In GEO forward-link MB-HTS, RZF mitigates rather than cancels mutual interference, and scheduling seeks strong channels and semi-orthogonality [2110.02525]. In return-link MF-TDMA, interference is addressed through path selection, MMSE-SIC ordering, and, when necessary, free-slot assignment [1210.4329]. In dynamic frequency assignment for mobile users, interference is largely represented through adjacency, reuse-group, and polarization constraints rather than explicit SINR optimization [2403.07007]. A common misconception is that MB-TFSA is merely a slot-labeling problem; in the cited works, the assignment is valuable only because it reshapes interference geometry.

The literature also reveals ambiguity in whether “frequency” must be explicit. One paper states that with reuse factor $1$, there is effectively a single frequency slot and MB-TFSA reduces to MB-TSA [2110.02525]. Another begins from dynamic frequency assignment and becomes MB-TFSA only after time discretization [2403.07007]. In the pilot domain, the “frequency” dimension is a phase-shift and support-separation mechanism rather than a user data channel [2505.04933]. This suggests that MB-TFSA is better interpreted as a generalized multi-beam resource-index assignment over a two-dimensional temporal-spectral structure, even when one dimension is degenerate or embedded in pilot design.

Complexity remains a central limitation. In OFDMA-SDMA, per-dual-iteration complexity is dominated by pseudo-inverses for each candidate SDMA set, yielding overall per-iteration complexity $O(NK^M M^3)$, though pseudo-inverses can be precomputed once per $(n,t)$ [1212.5284]. In the return-link scheduler, exhaustive search requires $M(M!)^{B-1}$ rate evaluations, while bipartite-path search reduces this to $M^B$ [1210.4329]. In pilot scheduling, DSatur grouping is $O(U^2)$ worst-case, and phase-grid search adds $O(C\cdot |\Phi|\cdot |\Phi_t|)$ [2505.04933]. In the QUBO formulation, total qubits equal the number of main decision variables plus slack qubits, and the slack-bit count can dominate without aggressive rescaling [2603.00701].

Finally, the reported limitations are domain-specific but structurally similar. ZF requires full column rank and $|S_{n,t}|\le M$ [1212.5284]. GEO static-operation assumptions rely on quasi-static channels and accurate gateway CSI [2110.02525]. Return-link MMSE-SIC ordering assumes perfect CSI [1210.4329]. The TFPSP separability theorem is asymptotic in large $M,K,N_p$, and finite grids or synchronization impairments may preclude perfect non-overlap [2505.04933]. The quantum formulation is currently restricted to very small instances because qubit counts and circuit depth remain the bottlenecks [2603.00701]. Across all cases, a plausible implication is that MB-TFSA research is less constrained by modeling expressiveness than by the computational and physical cost of enforcing coupled constraints at operational scale.

Source: https://www.emergentmind.com/topics/multi-beam-time-frequency-slot-assignment-mb-tfsa