Get the newsletter
Qualcomm · Networking · Filed May 7, 2025 · USPTO published Nov 12, 2026 · This week's release

Qualcomm Patents a Phone That Predicts Whether Its Upload Will Get Through

Qualcomm's patent teaches your phone to check recent signal conditions and guess whether its next upload will get through. If the odds look bad, the phone holds off.

Cell towers exchange wireless signals with phones, drones, and vehicles moving across a cellular network. Drawing from patent filing US 2026/0362431 A1.
Cell towers exchange wireless signals with phones, drones, and vehicles moving across a cellular network.All 13 drawings ↓
Publication number
US 2026/0362431 A1(US20260362431A1)
Applicant
QUALCOMM Incorporated
Filed · Published
May 7, 2025 · Nov 12, 2026
Inventors
Yi HUANG, Taesang YOO, Himanshu JOSHI, Ziqi WANG
Status when we published
Waiting for an examiner
Document
20 claims
US classification
370/329

What Qualcomm's upload-prediction patent does

You're walking through a crowded train station and your phone has a message to send. The network has set aside regular sending windows for your phone, and by default it grabs the very next one, even if the airwaves are jammed and the message will probably get lost.

Qualcomm's patent gives the phone a way to check first. Before each window, a small AI model looks at how the signal and interference behaved a moment ago, plus how your recent uploads and downloads went, and estimates the odds that sending now will work.

The network sets the bar for how good those odds must be. If your phone clears it, the message goes out. If not, your phone holds off and tries a later window instead of spending a send on a likely failure.

From the filing · CLAIM 1
receive information that identifies a threshold level corresponding to a probability of success for an autonomous physical uplink shared channel (PUSCH) transmission by the UE; …

Translation: The phone checks the likelihood that its data upload will succeed before sending it.

How the phone scores each upcoming send slot

The filing covers both ends of the link: the phone (called a user equipment, or UE) and the network equipment, such as a base station. The network first sends a threshold level, a minimum chance of success for an upload the phone starts on its own. It also sends a configuration listing one or more configured grant setups, each a set of pre-scheduled sending slots the phone may use without asking permission each time.

Before the next slot, the phone can run a learning model. Its inputs, drawn from the slots just before, can include:

  • Recent signal-quality measurements
  • Recent interference measurements
  • How often its recent uploads succeeded
  • How often recent downloads succeeded

If the model's output meets the threshold, the phone sends in the next slot. If not, it waits for a later one. The model can be a neural network, a CNN or an RNN (types of AI that spot patterns in a sequence of recent readings), and the description also covers a transformer.

When the network has set up several configurations with different transmission settings, such as how densely data is packed, the model can also choose which one to use, or choose to wait.

The network can help train the model by sharing compressed signal data, register the phone's models, switch them on or off, or tell the phone to swap between a more optimistic and a more pessimistic one. In the example, a network success rate above a 90% target suggests the model is too cautious.

From the filing · THE ABSTRACT
… user equipment (UE) autonomous uplink transmission based on artificial intelligence/machine learning (AI/ML) modeling.

Translation: The device uses machine learning to decide when and how to upload data independently.

Why a phone that waits could save you retries

The practical payoff is fewer wasted sends. When a phone fires off data into a noisy slot, the network can't read it, and the work has to be redone. A phone that skips those slots leaves its airtime for sends that have a better chance of landing, which the filing says should improve resource use and cut interference across the network.

The design also gives the carrier a dial. The filing describes a trade-off between delay and success rate: a more optimistic model sends sooner, a more pessimistic one fails less often. The network can pick which behavior it wants and change it as conditions shift, which matters for anything that sends data on its own, from phones to cars and industrial gear.

Qualcomm's 590th filing we have tracked since May adds to our Qualcomm coverage that includes car safety alerts and touch shortcuts for phones.

Source: US 2026/0362431 A1 at the USPTO (PDF). Full patent text and figures are in the official publication.

Get our take in your Top Stories

Liked this breakdown? Add Patentlyze as a preferred source on Google, and our plain-English take shows up more often in your Top Stories the next time Qualcomm patent news breaks.

There are more where this came from

We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.

The drawings · 13 sheets

13 drawing sheets from US 2026/0362431 A1 · click any drawing to enlarge

Cell towers connecting to phones, cars, and drones, drawing sheet 1 from US 2026/0362431 A1
Sheet 1 Cell towers connecting to phones, cars, and drones
An AI predicting whether a phone's next upload succeeds, drawing sheet 2 from US 2026/0362431 A1
Sheet 2 An AI predicting whether a phone's next upload succeeds
Two alternative neural networks that forecast upload success, drawing sheet 3 from US 2026/0362431 A1
Sheet 3 Two alternative neural networks that forecast upload success
An attention-based neural network that predicts upload delivery, drawing sheet 4 from US 2026/0362431 A1
Sheet 4 An attention-based neural network that predicts upload delivery
Using AI predictions to pick upload settings or wait, drawing sheet 5 from US 2026/0362431 A1
Sheet 5 Using AI predictions to pick upload settings or wait
Calculating odds for every upload format before transmitting, drawing sheet 6 from US 2026/0362431 A1
Sheet 6 Calculating odds for every upload format before transmitting
Comparing separate AI forecasts to make the final upload call, drawing sheet 7 from US 2026/0362431 A1
Sheet 7 Comparing separate AI forecasts to make the final upload call
Internal software components that manage the predictive uploads, drawing sheet 8 from US 2026/0362431 A1
Sheet 8 Internal software components that manage the predictive uploads
Show the other 5 sheets
Hardware inside the phone that handles AI-guided uploads, drawing sheet 9 from US 2026/0362431 A1
Sheet 9 Hardware inside the phone that handles AI-guided uploads
Internal controls cell towers use to guide AI uploads, drawing sheet 10 from US 2026/0362431 A1
Sheet 10 Internal controls cell towers use to guide AI uploads
Components in a cell tower managing uploads from phones, drawing sheet 11 from US 2026/0362431 A1
Sheet 11 Components in a cell tower managing uploads from phones
How the phone decides when to send an upload, drawing sheet 12 from US 2026/0362431 A1
Sheet 12 How the phone decides when to send an upload
Steps the cell tower takes to set upload rules, drawing sheet 13 from US 2026/0362431 A1
Sheet 13 Steps the cell tower takes to set upload rules
Patent filing page

Reader comments

No comments yet

Start the discussion

Real name or a handle, either is fine. Comments are read by a person before they appear, so allow a little time. Keep it about the filing.