Startup claims there's much more data to mine from cable's legacy upstream

(Source: Science Photo Library/Alamy Stock Photo)

Cable's DOCSIS network has always been asymmetrical – heavy on the downstream and light on the upstream. That asymmetric nature has been driven by how much spectrum has been allocated in each direction, and has largely followed broadband usage patterns as downstream data continues to outrun the upstream by a wide margin.

While that downstream-to-upstream usage disparity remains, usage trends in recent years clearly indicate that growth is tilting toward the usptream. OpenVault, which publishes a quarterly "Broadband Insights" report that analyzes anonymized data from broadband operators, found in its Q2 2026 study (registration required) that average upstream usage on DOCSIS networks grew 19.2% to 58.2 gigabytes in the period, while the rate of downstream usage declined by 6.1%.

Upstream action

That is also leading to an adjustment in traffic ratios. In Q2 2026, the ratio was 11.2:1 in favor of the downstream, from a ratio of 16.6:1 in 2019, according to OpenVault data.

Related:The nature of upstream network usage has likely changed forever, OpenVault says

(Source: OpenVault Q2 2026 Broadband Insights report)

OpenVault holds that this shift toward more upstream usage – being fueled in part by video conferencing and devices performing cloud syncs – is "not a seasonable blip." Instead, OpenVault views it as a clear "structural" indicator that cable operators need to prioritize improving upstream speeds and capacity to stay ahead of growing demand.

And many of them are. While a contingent of cable operators is deploying more efficient upstreams based on blocks of OFDMA (Orthogonal Frequency-Division Multiple Access) subcarriers that were introduced with DOCSIS 3.1, others are dedicating more spectrum to the upstream via "mid-split" and "high-split" upgrades and have begun the push toward DOCSIS 4.0.

Some of that activity is showing up in the data: OpenVault found that average upstream speeds rose 33.7% to 47.3 Mbit/s in Q2.

'Like a virtual node split in the upstream'

But those options require some work, including changes to the network, and can't be executed overnight.

A startup called ANSpect Technologies claims to have developed technology that provides an easier and faster way for cable operators to managing this emerging pain point by mining more bits out of their legacy ATDMA (Advanced Time Division Multiple Access) upstream channels that operate in the low, noisy portion of the spectrum.

The general idea posed by ANSpect is to run proprietary channels alongside the legacy ATDMA channels that operate today in a 5MHz-42MHz swath of spectrum known as the "low-split" on a North American DOCSIS network (the low-split for a EuroDOCSIS network runs at 5MHz-65MHz). The startup claims this approach can provide operators with an additional 35 Mbit/s to 80 Mbit/s of upstream capacity in this part of the spectrum. That roughly equates to a 35% to 85% increase in throughput in the upstream when compared to an ATDMA-only upstream.

Related:OpenVault CEO: As uplink usage keeps rising, cable operators will be unable to rest

This chart illustrates where ANSpect's co-channel would run alongside legacy ATDMA channels in the lower part of the cable spectrum. ANSpect has also adapted its technology to mine more data out of the cable downstream, but it stressed it is initially focused on enhancing the cable upstream. (Source: ANSpect Technologies)

"We're taking old spectrum and getting new bits out of it," John Chapman, ANSpect's SVP and chief strategy officer, cable, explained. "It's almost like a virtual node split in the upstream."

Chapman, a DOCSIS pioneer and former longtime Cisco Systems exec, said the true sweet spot for this approach is to nearly double the upstream capacity for DOCSIS 3.0 and D3.1 modems that speak ATDMA, a technology that's been in use by cable for about three decades.

"I would submit that ATDMA under-utilizes the spectrum," Chapman said. "We can make that DOCSIS 3.0 and DOCSIS 3.1 service in the upstream work better. … We mine, out of thin air, the unused bits that ATDMA has not already mined out of there. It's like we're dressing up ATDMA to act like OFDMA."

Related:Clock ticks for cable ops as Cisco's flagship CMTS nears end-of-life

Deployment scenarios under evaluation

ANSpect declined to detail the secret sauce in its technology and how it can create new channels that can coexist with ATDMA in the same spectrum block. But it demonstrated its handiwork at the recent SCTE Tech Expo in Atlanta. There, it emulated a cable modem with Calian test equipment and then emulated how can inject its channels to run alongside the legacy ATDMA channels.

ANSpect, which intends to license its wares to cable network suppliers, envisions its technology being deployed on hybrid fiber/coax (HFC) networks that have employed a virtual cable modem termination system (vCMTS) in tandem with a remote PHY device deployed at the node in a distributed access architecture (DAA) or a remote PHY shelf installed at the headend.

The company is evaluating whether its system can be deployed via a software upgrade at the RPD or RPHY shelf or if it will require some sort of hardware add-on. Its technology is made to work with existing DOCSIS modems.

While there's a clearer path to a software-only deployment for RPDs that use FPGAs (field-programmable gate arrays), ANSpect is looking into a secondary approach for deployments involving RPDs that run on ASICs (application-specific integrated circuits), explained ANSpect co-founder and CTO Dr. Ahsan Naim.

Chapman said the company has "penciled out ways" to run its technology on older integrated CMTSs, but he said the primary focus is on the vCMTS. Cable operators, he points out, are pivoting toward virtual CMTSs as support for legacy i-CMTSs starts to wane. Notably, Cisco's former CMTS workhorse, the cBR-8, has entered the end-of-life process. "We're looking to ride that [end-of-life] wave," Chapman said.

Targeting field trials in 2027

Chapman noted that ANSpect's technology can also be adapted to mine more bits – perhaps up to 2 Gbit/s more – out of the DOCSIS downstream. But the company's emphasis heading into 2027 is on the cable upstream.

"Upstream is where the pain point is" for cable, he said.

ANSpect is aiming to get into cable field trials in 2027, and discussions are already underway with tech partners and operators.

The speed at which those trials will occur will depend on how the deployment scenarios pan out. While a software-only approach would put the company in position to enter trials in the early part of 2027, both options are "doable" in 2027, said company co-founder and CEO Surya Panditi.

Agnostic approach

Founded in late 2025, ANSpect believes its technology can be applied to multiple types of networks, including HFC, fixed wireless access (FWA), cellular and satellite.

On the mobile side, the company is participating in the Commonwealth Cyber Initiative's xG Testbed at Virginia Tech, an initiative that includes AT&T, Verizon, Samsung and SoftBank among its participants. The results of a recent test of indoor and outdoor applications involving the company's O-EMA (Orthogonal-Enhanced Multiple Access) technology designed to extract more capacity from existing wireless networks could be shared before the end of the month, according to Panditi.

The startup, which is in the pre-revenue stage, currently has seven full-time employees and has raised an undisclosed amount of funds via a pre-seed round. ANSpect is considering a potential follow-on funding round after obtaining some interest from a pair of mobile technology suppliers, Panditi said.

Other ANSpect employees and advisors are well known in cable tech circles, including Daniel Etman, the company's head of business development for Europe (late of Cisco, Harmonic and Casa Systems); Luca Martini, SVP of systems engineering (late of Cisco and Charter Communications); and advisors Ken Klaer (formerly with Scientific-Atlanta and Comcast) and Sandy Howe (formerly with Technetix, Arris and Cisco).

About the Author