Blog


Managing Wi-Fi Across Legacy CPEs with Incognito

Kevin Mukai
Director Product Marketing, Wi-Fi Solutions

Posted on August 8, 2018

On July 31 ASSIA and Incognito issued a joint press release with the headline:  Incognito and ASSIA Announce Partnership for Proactive Wi-Fi Management.   ASSIA has a complete set of broadband and Wi-Fi management solutions for ISPs and MSOs.  This leaves one to wonder about the importance and value of such a partnership?

In order to better understand the value of such a partnership, let’s delve deeper into the technology benefits of ASSIA’s CloudCheck for Wi-Fi management, and let’s further consider what Incognito brings to the table for service providers.

Challenges of Deploying Wi-Fi Management

Today’s global service providers are faced with an ever-increasing barrage of Wi-Fi related customer care calls.  Operators throughout the world have shared with ASSIA that Wi-Fi problems account for 50% or more of their in-bound customer support inquires.  This creates a heavy resource burden and drives up OPEX for customer care.

The ever-increasing demand on home Wi-Fi networks requires sophisticated and effective solutions that provide visibility and proactively resolve issues through automated optimization across the service provider’s entire subscriber network.   The dynamic nature of Wi-Fi usage across a multitude of devices in the home requires a management solution to operate in real-time and provide intelligent proactive care.

Service providers also want to provide Wi-Fi management across a mix of CPE types, including legacy CPEs and newer models that are deployed in the network.  They are looking for a solution that can address the entire network, without requiring massive upgrades in hardware.

ASSIA has found service providers need Wi-Fi management solutions that offer:

  • Advanced, intelligent, software capabilities that incorporates deep diagnostics, analytics, and learning-base optimization
  • Upgrades and scales across all CPE types as the subscriber network evolves
  • Is simple and easy-to-use, yet flexible , for call centers, agents, and field technicians

Solving for the Impossible?

Vendors in the market today provide a variety of different approaches to address Wi-Fi management.  These approaches include gateway/hardware-based solutions or ACS/TR-069 based methodologies.  While beneficial in some ways, there are shortcomings to these approaches.

Gateway/HW based solutions lack adequate the local compute power and storage to handle sophisticated diagnostics and contextual-based optimization.   Any Wi-Fi management capability is tied to specific CPE models only.

ACS/TR-069 based solutions lack the real-time performance and rich diagnostics and optimization capabilities required for managing residential Wi-Fi today.  Addressing the dynamic performance requirements across millions of homes requires a system that can efficiently scale and optimize Wi-Fi networks across all use cases.

Unlike hardware and TR-069 based solutions, the CloudCheck cloud and agent-based architecture provides the real-time control and management coupled with the historical, machine-learning based optimization.  The CloudCheck software solution allows deployment across all CPEs in the service provider’s entire network.

Benefits of the Incognito Partnership

Incognito offers an ACS system which provides provisioning capability for CPEs across all CPE makes and models based on TR-069 and TR-181.    The joint ASSIA + Incognito solution integrates the ASSIA CloudCheck solution with the Incognito Digital Service Platform, extending a portion of the CloudCheck capabilities across legacy, agent-less devices.  In addition, the integration solution provides a unified customer care interface across both legacy and agent-based CPEs.

Until now, legacy devices have had very limited Wi-Fi management capability.  By extending the CloudCheck solution to the Incognito platform, an operator can have a unified customer care interface to support a subset of CloudCheck capabilities across legacy CPEs as well a comprehensive suite of diagnostics and optimization features across modern CPEs with CloudCheck agents.  This creates a more uniform and consistent care experience across all classes of devices in a service provider’s network.


Deployment of New xDSL Technologies, Challenges and Opportunities

Mehdi Mohseni
Chief Scientist

Posted on July 25, 2018

Deployment of new xDSL technologies, including vectored VDSL or G.vector, Super Vectoring or VDSL2 35b and G.fast is a significant opportunity for broadband service providers as it opens many opportunities for offering new services and enhanced revenue streams. Vectored VDSL and Super Vectoring can deliver rates in the order of 250 Mbps while G.fast is targeting 1Gbps at a significantly lower cost compared to fiber-to-the-home (FTTH) installations. A number of studies suggested that the deployment cost is lower at least by a factor of five for vectored VDSL and by a factor of two for G.fast. Although the deployment cost of these new xDSL technologies is estimated to be substantially lower relative to FTTH, it still represents a major investment, which service providers need to carefully plan, deploy and operate in order to maximize their return on investment (ROI).

Service providers upgrading their access network to vectored VDSL or G.fast are faced with various challenges for achieving their desired ROI. The main challenges can be summarized briefly as in the following:

  1. Service qualification: Service providers require reliable projections for data rates that can be delivered for these new services. They rely primarily on attenuation-based qualification rules that predict the achievable rates based on loop attenuation inventory. However, this approach does not take into account the effect of non-crosstalk (alien) noises, physical loop impairments, and inaccuracy of loop inventory data on accuracy of the predictions. As a result, such attenuation-based qualification rules usually have high false positive rates (sunk costs), or if designed conservatively, they would suffer from large false negative rates (lost opportunities).
  2. Outside Plant “grooming”: Service providers require identifying and resolving pre-existing line faults that potentially limit or deteriorate the expected performance gain of these new technologies.
  3. Greater needs for stabilization: Vectoring removes Far-End Crosstalk (FEXT) among the vectored group of lines exposing them to alien and impulsive noise sources that were hidden under FEXT before vectoring. Moreover, G.fast employs a much wider bandwidth compared to VDSL2. This makes G.fast exposed to a bigger set of interference and impulse noise sources that can have serious detrimental impact on the service stability. These additional noises need to be managed to avoid any detrimental impact on the stability of the lines.

DSL Expresse (DSLE) Module Summary

ASSIA’s DSL Expresse™ offers the following various tools and modules to help address these challenges:

  1. Service Recommender and Predictor Module: ASSIA’s Service Recommender Module estimates the viable rates and services that can be offered reliably with new xDSL technologies. With potential shortcomings of attenuation-based qualification rules in predicting services with vectored VDSL and G.fast, ASSIA’s Predictor Module is a reliable tool required to enable the service providers to operate the new lines at the maximum possible rates stably.
  2. Diagnostics module: This diagnostics tool helps to identify loop faults and provide precise guidelines on maintenance actions needed in the outside plant and in-house wiring to enable these new technologies. Loop impairments limit gains from these vectored technologies. Because these technologies transmit over a larger bandwidth and also, they use vectoring to remove crosstalk, the impact of loop impairments is significantly higher on the performance compared to ADSL or VDSL. Moreover, with customer self-installation of CPE equipment, various in-house wiring issues are not cleared and would potentially deteriorate the performance, significantly in some cases; thus, there is a need to identify and resolve wiring and/or noise issues to maximize the gains offered by vectored services. ASSIA’s Diagnostics Module is also capable of analyzing the crosstalk coupling measurements among the vectored group of lines (in addition to DELT data) to identify faults and issues that would impact the performance. The crosstalk analysis provides additional guidelines for cleaning up the faults and wiring issues.
  3. Profile Optimization: The Profile Optimization Module manages and improves the stability. With potential exposure to other sources of non-crosstalk noise, there is a need for this profile optimization tool to monitor and if needed to address the impact of such alien and impulsive noises and avoid any detrimental impact on the stability and performance of the new services. Moreover, as lines are pushed toward their limits in these new deployments, the stability issues become more prominent and would negatively impact the experience of the customer if not properly addressed. G.fast employs G.inp physical-layer retransmission mechanism as standard for offering a more effective protection against impulsive noises. For vectored VDSL, G.inp is an option. ASSIA strongly recommends deploying vectored VDSL together with G.inp physical-layer retransmission mechanism and optimizing the configurations of G.inp for these new deployments to improve the stability of the lines. Unmanaged use of G.inp for both cases may hide performance issues (e.g., throughput drops) that would impact the user experience negatively.

Although any upgrade to a new advanced technology can be challenging and costly even when it leverages already installed copper lines, the use of tools like the modules outlined above available by ASSIA, can make the transmission much smoother while also yielding a higher return on investment.


Network Management and Network Optimization

Sina Zahedi
Senior Director, Systems Engineering

Posted on July 11, 2018

Network management and network optimization are critical elements of internet service provider (ISP) operations. The following offers insight in these functions.

Network management is a set of processes and practices used by service providers and enterprise network managers to administer and manage networks. Network management touches every aspect of the network operation and often includes several well-defined and highly optimized set of workflows.  These workflows vary depending on the type of the services the ISP offer, and the range of technologies and equipment involved for enabling the offering. These functions usually can be categorized in the following groups:

Provisioning:

This includes all activities required for initial setup of service to the customer. There is a high degree of dependence to the type of service provided. For example, for a fixed line access network, this could include:

  • running copper or fiber to the customer premises
  • setting up the equipment
  • creating accounts on relevant internal IT systems for authentication and administration
  • setting up the billing system, building proper data routes
  • certifying that the service is established and is operational

The workflows for these processes are often well defined and highly automated at large ISPs.

Despite this high degree of automation, certain aspects of this process could become quite challenging for operators. For example, deciding on the proper range of service options for customers could be quite challenging in situations where the transmission technology imposes limits due to distance, quality of transmission lines or noise characteristics of the environment. While an operator’s highly automated processes are designed to handle a large majority of cases efficiently, a small number of customers may experience long delays and failed attempts at establishing their service. They may also experience poor quality at the beginning of their subscription with the operator or after service upgrade to higher rate services. While operators may choose to employ a conservative approach to minimize the likelihood of such incidents, they may lose business advantage to their more aggressive competitors or fail to monetize their investment to the full extent possible by refusing to offer higher priced services that the customer may purchase.

Monitoring:

This generally includes collection, storage and analysis of a large variety of data from every element in operator’s network. The goal of this process is to ensure continuity of service, monitor the health of systems, identify and address large scale issues – as well as to provide visibility to network managers for troubleshooting. Network operation centers (NOC) generally collect and monitor a wide selection of network parameters in different segments of the network. This allows the providers to ensure critical network infrastructure equipment operate as intended and generate proper alarms and notifications when equipment malfunctions or are under excessive load pressure.

Monitoring also enables tracking of key network KPIs and allows network managers to identify weak areas of the network that could benefit from equipment upgrades and new investment. SNMP protocol is commonly used to interface with different network elements and is widely used for collecting data from equipment. Collected data is frequently used to generate alarms and notifications for corresponding organizations in charge of different segments of the network.

Fault Management and Customer Care:

Issues arise in different parts of the network – and depending on their scope, they could affect a single customer, a particular geographical area, or impact vast segments of the network. Identifying the correct scope of the problem and affected areas is a key aspect of the troubleshooting process. An issue on the core network is expected to affect several customers and there is a high likelihood the operator will receive many complaints from affected customers. To address these issues properly, the provider should be able to identify the issue as a network outage problem rather than trying to individually troubleshoot affected customers. Network outage issues are generally identified quickly as relevant equipment is constantly monitored and redundancy is usually built into core network functions. This approach reduces the likelihood of total network failures. However, identifying and troubleshooting issues that are affecting individual customers would likely be far more difficult. There are multiple reasons for this. First, the nature of issues affecting customer’s physical layer performance could be very varied and time-varying as well. Second, several different problems could have similar impact on experience of the customer and it is difficult to pinpoint the exact root cause of the problem simply based on customer’s comments. Third, an operator’s visibility to customer premises equipment is fairly limited and this makes monitoring and troubleshooting processes more difficult. In environments where the network operator and physical plant owner are different companies, additional issues arise as two where the fault is located and who is responsible for addressing it.

Considering that the vast majority of issues reported to operators stem from customers’ environments, establishing proper procedures for identifying and addressing these issues is of utmost importance. Unfortunately, it is common that the issue cannot be remotely identified or addressed and requires a technician visit to customer’s home. This is costly and frequently fails to appropriately identify and address the underlying issue due to lack of proper data, especially if the problem is transient in nature.

Network Optimization

Network optimization refers to a set of practices to ensure proper operation of the network at different times and in response to time varying changes. Some of these practices are static in nature. For example, a growing network area with higher traffic demand may require a stronger backbone network to handle the growing traffic – which may require investment in new equipment and transmission lines. Some other practices require dynamically adjusting and tuning some network parameters in response to network’s changing behavior. For example, Wi-Fi routers interfering with a customer’s Wi-Fi network frequently change and may require adjustments to the channel or band the customer is using. Another example is a change in the level and characteristics of noise on a customer’s DSL line which could be due to a change in electromagnetic interference or change in the cross-talk level due to the addition of new customers in the same binder or cable serving the customer. These time variations make the network a dynamic environment and will require constant monitoring and frequent tuning in response to these changes.

 

ASSIA’s Approach to the Problem

ASSIA’s products strive to address existing gaps in areas of network management and network optimization. While core backhaul networks have enjoyed from a strong and robust set of tools for monitoring and optimization over the past decade, the access side of the network has not seen a robust set of monitoring and optimization tools. This is even though the access environment is very complex due to the variety of different access technologies and the diversity of scenarios that could affect the customer’s experience of service.

ASSIA’s products employ a deep understanding of the physical layer of different access technologies, including DSL, Fiber and Wi-Fi, and employ an array of statistical analysis to identify issues and perform actions to alleviate the impact of those issues without human intervention. ASSIA’s DSL Expresse, GPON Expresse and CloudCheck products collect and maintain a large set of data and analyze variation of these parameters over time to identify issues that may not be present or visible at a particular instance of time. Such analysis is not in the purview of chipsets or communication equipment due to the required memory and processing power. Further, collected data and time analysis of collected data allows operators to use appropriate data in various work flows allowing improved performance and data-driven decision-making.

While network optimization remains a critical part of ASSIA’s products, these software solutions are also progressively becoming a more and more critical part of the network management suite as operators find creative use cases for the data and analytics results. Today, much of the provisioning of many DSL operators are based on the service recommendations of DSL Expresse, and many operators use DSL Expresse, GPON Expresse and CloudCheck for monitoring their network and troubleshooting their reported customer issues. While physical layer optimization continues to remain a critical function of ASSIA’s software suite, network management functions over the years have become equally important and central to operation and success of many of ASSIA’s customers.

 


Quality of Experience With CloudCheck: Perception vs. Reality

Kamal Yassin
Director of Product Management

Posted on June 27, 2018

Senior executives of 362 companies from different industries were recently asked about their customer experience.  As one might expect, 80% of them said that their company provides an outstanding customer experience.  Unfortunately, only 8% of their customers concurred with those positive assessments [1]. The executives’ responses were based on service quality metrics they use to gauge  the satisfaction level of their subscribers.  This disparity between the executive perspective and customer view of experience highlights the difference between assessments based on the service quality versus actual customers’ perception of their experience.

The Importance of Quality-of-Experience

Because of such discrepancies in customer perception of service quality, Quality-of-Experience (QoE) has become one of the most important indicators that measure the customers’ actual satisfaction with their service.  An emphasis on QoE has become one of the key differentiators in the marketplace between service providers in this socially-connected era.   This is important to service providers as high satisfaction equates to higher customer retention as well as lower operational expenses, not to mention the marketing and the positive brand perception promoted by vocal customers.

 

What About Quality-of-Service (QoS)?

QoS is an important metric that measures the level of the service.   As defined by one source that defines QoS in the telecommunications and computer networking arena, QoS “refers to traffic prioritization and resource reservation control mechanisms rather than the achieved service quality. Quality of service is the ability to provide different priority to different applications, users, or data flows, or to guarantee a certain level of performance to a data flow[1]. Many technical knobs exist in the Customer Premise Equipment (CPE) to manage access and control the performance of the different stations inside the home. QoS is focused on evaluating these controls, verifying their functionality and performance while adjusting priorities and resources to guarantee the IPTV performance or the VoIP quality, for example.

Computing QoE

QoS is one of the components that is used to measure and improve QoE.  However, QoE could be very subjective and not necessarily linked directly to the quality of the service rendered. QoE represents the aggregation of the “total experience” of the individual customers and their perception of their service.  This could be affected by uncontrolled or environmental factors.  In residential networks, QoE is an end-to-end measurement that includes both the broadband and Wi-Fi.  It is evaluated using customer data such as calls, customer surveys, NPS, and churn among other factors.   Hence, computing QoE is challenging because QoE is:

  • Not quantifiable
  • Subjective
  • Inconsistent
  • Expensive
  • Hard to track
  • Reactive

To solve the QoE puzzle, one must:

  • Identify technical parameters that relate to the customer experience
  • Quantify
  • Measure as a whole, but pinpoint the root cause
  • Correlate
  • Calibrate

In a previous blog, ASSIA discussed the importance of measuring throughput.  The first step to quantifying QoE lies in measuring and separating the broadband throughput from the Wi-Fi throughput for every station connected to a given Wi-Fi network.

While throughput is the most relevant technical parameter to QoE, it is not the only one. Connectivity and latency combined with throughput provide greater insight into a subscriber’s sense of their connection and perception of their experience.   However, such metrics can also be adversely impacted by external factors related to coverage, interference, congestion, and hardware configuration. Incorporating a multi-variant model that combines these different measurements and diagnostics to generate a single QoE score that measures the customer experience offers tremendous benefit.  ASSIA’s CloudCheck provides an overall QoE score in addition to quality scores for both Wi-Fi and broadband.

CloudCheck QoE Index

Once this quantifiable index for QoE is calculated, the model is continually tuned and validated.

CloudCheck provides a machine learning engine in the cloud server that uses external data sources such as call data, customer surveys, dispatch records to:

  • Correlate QoE with customer input data to continually validate the QoE score.
  • Calibrate periodically to adjust for seasonality and other changes in the network over time.
  • Categorize performance levels such as turning the numbers into a Red, Amber, Green (RAG) index.

Learning Engine

CloudCheck’s QoE Index and its machine learning engine offers several benefits to the service providers:

  • Quantifies QoE into a single technical automatically-generated metric.
  • Incorporates periodic and automated calibration to maintain accuracy.
  • Uses a predictive model that tracks the performance trends and signals potential customer dissatisfaction at the network level which would eventually translate into more complaints and higher churn.
  • Presents a similar measurement per individual line and links any degradation to its main driver(s).

The same approach used to conceive the QoE score can be used to derive more specialized quality scores that focus on specific services such as IPTV, VoIP, channels (Wi-Fi vs. broadband), and hardware (gateways, mesh, APs, extenders …). In a future blog, we will look more closely at these quality scores and their use cases.

[1] http://www.bain.com/publications/articles/how-some-banks-turn-clients-into-advocates.aspx

[2] https://en.wikipedia.org/wiki/Quality_of_service#End-to-end_quality_of_service


Why Cloud for Wi-Fi Management?

Jarrett Miller
Vice President of Global Business Development

Posted on June 20, 2018

Today’s global service providers are faced with an ever increasing barrage of Wi-Fi related customer care calls.  Operators throughout the world have shared with ASSIA that Wi-Fi problems account for 50% or more of their in-bound customer support inquires.  With an overall objective to significantly reduce the number of these costly customer care engagements while providing visibility and control over the balance, operators require a sophisticated Cloud/Agent based Wi-Fi management solution.

In recent years, operators have come to realize that Cloud/Agent Based Wi-Fi management platforms – such as ASSIA’s CloudCheck solution are required and are vastly superior to TR-069 or other and gateway-based (SON) solutions.  Below we will discuss why the market is embracing this Cloud/Agent architecture.

The Cloud and Wi-Fi Management

The ever increasing demand on Wi-Fi networks in the home today require sophisticated and capable solutions that provide visibility and resolve issues through automated optimization.  Gateway devices simply cannot offer sufficient compute power to perform analysis, nor dedicate nearly enough storage space on board to facilitate optimization.  Such a system requires an agent on the gateway working in harmony with an engine running complex algorithms in the cloud.

CloudCheck was developed as an Cloud/Agent solution from its inception.  This architecture serves as a foundational strength of CloudCheck and enables the solution to excel at Wi-Fi management to some of the world’s most progressive service providers.

Cloud Compute Power

The Cloud – where ASSIA’s CloudCheck Server resides – offers unlimited storage and unlimited compute power.  CloudCheck needs this to:

  1. Collect extensive detail on each subscriber gateways and the numerous laptops, tablets, smart phones, and other nodes connecting to those gateways
  2. Process detail from the CloudCheck agents on gateways and the nodes
  3. Develop histograms for each subscriber gateway
  4. Run numerous CloudCheck algorithms to make recommendations and create an ideal policy to manage each Wi-Fi environment
  5. Enable real-time processing of information
  6. Perform historical analysis
  7. Determine recommendations
  8. Establish optimization actions for each gateway and node in the environment
  9. Calculate Quality of Experience Score
  10. Create user, neighborhood, regional or network-wide reporting

Sophistication & Scale

CloudCheck runs a number of sophisticated algorithms to enable the Wi-Fi management platform.  These algorithms are continually enhanced and require a large amount of compute power.  Operators need the ability to manage on a single subscriber, neighborhood, city, region or a global basis.  The ability of the cloud to house significant amounts historical data for a broad number of subscribers is key – as it makes the data available for analysis.

The more information CloudCheck’s algorithm is fed – the better the result and the greater the accuracy.

Context & Machine Learning

Home Wi-Fi networks are highly dynamic.  The ability to make intelligent recommendations and effective optimization actions are essential for operators to maintain a high quality of experience (QoE) for subscribers.

CloudCheck performs complex real time analysis and compares its findings to the knowledge gained through historical analysis.  CloudCheck gathers and analyzes histogram data across every station on each band of the Wi-Fi gateway.  It juxtaposes this against real-time usage information it receives from the gateway to determines context.  This notion of context allows CloudCheck to be highly effective and extremely accurate in its optimizations.

In our previous blog series,  ASSIA’s Approach to Machine Learning, we discussed how CloudCheck exploits machine learning to improve Wi-Fi Quality of Experience.  The cloud has the scale and power to run AI and machine learning algorithms efficiently while improving the accuracy and effectiveness of optimizations over time.

In comparison, context is simply not possible in a TR-69 or gateway based SON solution.

Visibility

Although the overall intent is to minimize support calls through optimization, it is essential for operators to have visibility in to subscriber home Wi-Fi environments in order to react to the remaining problems, proactively address issues, and positively impact the Quality of Experience (QoE) of each of their subscribers.  A Cloud/Agent-Based management system is essential in giving real time visibility and historical insight in to a home Wi-Fi environment.

CloudCheck provides real-time visibility to key elements of a subscriber Wi-Fi environment, such as:

  1. Network Up/Down
  2. Device health
  3. CPU Utilization
  4. Broadband speed
  5. Wi-Fi speed for every node in the environment
  6. Deep Analytics of the Wi-Fi environment

Because CloudCheck houses its data in the cloud, it can be leveraged to provide visibility to operator call centers through an ASSIA GUI or through the operator’s existing OSS utilizing CloudCheck APIs.  In addition, a CloudCheck app (or operator enabled custom app) can provide visibility to the home Wi-Fi environment for Operator Field Technicians and Subscribers alike.


ASSIA’s Approach to Machine Learning

Tuncay Cil and Haleema Mehmood
Team ASSIA

Posted on June 13, 2018

Part 2: Machine Learning improves Wi-Fi Quality of Experience

In the previous Machine Learning Primer article (published as a blog on ASSIA’s web site on May 30, 2018), we tried to give a short overview of basic AI/Machine learning definitions and the associated challenges.  In this article, we aim to provide some real-world examples of how ASSIA applies these AI/machine learning concepts on its CloudCheck® product line.

CloudCheck uses AI and machine learning techniques to solve a wide range of Wi-Fi related operational problems including AP (Access Point) level diagnostics, optimization, and network-wide trends and predictions.

AI and Machine Learning Examples:

Here we outline a step-by step process that tracks two example applications – Client Steering and Temperature Detection –  at the AP level:

  • Measure Health, Environment and Performance: CloudCheck measures the AP’s health, its environment and the performance of connected clients periodically. The period of observation varies by the type of data being collected, ensuring sufficient coverage of the observed phenomenon. For example, the RSSI of connected clients, which can show high variation with time, is measured at a small time-scale, while the temperature of the CPU, which remains largely constant for long time periods, is collected at a larger time-scale.
  • Diagnose: CloudCheck analyzes the data to determine trends and patterns at various levels. For example, for each client, a mapping of RSSI and throughput is established in both 2.4GHz and 5GHz frequency bands. In contrast, the temperature measurement is aggregated by the device model of the AP to determine the temperature distributions of different chipsets.
  • Set Thresholds and Policies: CloudCheck uses the detected patterns to establish thresholds or policies for diagnostics and optimization. For example, for each client, CloudCheck individually determines the best RSSI to transition from the 5GHz to 2.4GHz band, and the best RSSI to transition from the 2.4GHz to 5GHz band, optimizing for both throughput and coverage. For CPU temperature distributions, CloudCheck determines thresholds for finding outlying or problem temperatures.
  • Monitor Deviations: CloudCheck uses these policies to monitor deviations from the defined behavior. For Wi-Fi clients, any RSSI transition that crosses the established threshold for the particular band results in a real-time alert. For temperature distribution, daily temperature data collected for each AP is compared with the appropriate model-specific threshold.
  • Recommend or Optimize: The last step of the process is either an optimization action, or a diagnostic output combined with a trouble-shooting recommendation. Just like the period of observation, the output/response time is in accordance with the need of the application. In the client RSSI example, clients requiring band-steering are steered in near real-time to the appropriate band. For temperature, alerts are raised on a daily basis for each AP if required.

As more data is collected, CloudCheck is able to learn from the incoming data, improving the performance of the diagnostics and optimization actions over time.

QoE Assessment

Another application of machine learning techniques within CloudCheck is the prediction of user Quality of Experience (QoE). A supervised learning approach is employed for this purpose, using network-wide statistics:

  • Training data: Supervised learning for QoE requires a labeled data set for training. The features in this data come from CloudCheck’s daily AP diagnostics. The diagnostics are assessments of AP performance along individual dimensions/features including but not limited to client coverage and Wi-Fi throughput, interference in various bands, broadband disconnections and throughput, etc. The labels for the training data come from an assessment of user experience either based on customer surveys, or customer call/churn data provided by the service provider.
  • QoE Model Training: Supervised machine learning techniques are used on this training data-set to build a model that maps CloudCheck diagnostics to user QoE. This diagnostic is applied for daily assessment of customer QoE.
  • QoE Determined: Based on the needs of the Service Provider, this model can be used either for predictive purposes or for trouble-shooting. CloudCheck further employs feature selection techniques on the training data-set to identify the key drivers of customer dissatisfaction in the service provider’s network.

Over its 15 year history, ASSIA has pioneered development of AI/machine learning applications for last-mile and home network management challenges.  From their inception, ASSIA’s DSL, GPON and Wi-Fi solutions have utilized AI and machine learning to enable many of the world’s most progressive carriers and service providers to manage well over a hundred million internet connections.  Stay tuned for more innovation in AI at ASSIA in the future!


Net Neutrality: Accurate Speed and Quality of Experience Assessments Are Prerequisites

Djamel Bousaber
Sales Director

Posted on June 6, 2018

An evolution of the European regulatory framework

While the US Federal Communications Commission (FCC) voted to repeal Net Neutrality rules in December 2017, the European Union remains strongly committed to them. The adoption of the European regulation (EU)2015/2120 enforces a new fundamental right for citizens of the European Economic Area: a non-discriminatory access to an open Internet.

In addition to the policies for “Safeguarding of open access”, the regulation introduces the quality experienced by the end user as the baseline and states in paragraph 1 of Article 4 (point d and e):

Providers of internet access services shall ensure that any contract which includes internet access services specifies at least the following:

  • a clear and comprehensible explanation of the minimum, normally available, maximum and advertised download and upload speed of the internet access services in the case of fixed networks, or of the estimated maximum and advertised download and upload speed of the internet access services in the case of mobile networks, and how significant deviations from the respective advertised download and upload speeds could impact the exercise of the end-users’ rights laid down in Article 3;
  • a clear and comprehensible explanation of the remedies available to the consumer in accordance with national law in the event of any continuous or regularly recurring discrepancy between the actual performance of the internet access service regarding speed or other quality of service parameters and the performance indicated in accordance with points (a) to (d).

Dura lex, sed lex.

A challenge for regulators and Internet Service providers

Open Internet Word CloudNational regulatory agencies have the critical tasks to translate, enforce, and monitor an EU-compliant national implementation. This is leading to different initiatives like the sourcing of a reference system for Net Neutrality Measurement Tool by the Body of European Regulators for Electronic Communications (BEREC), a new Voluntary Code of Conduct released by UK regulator Ofcom in March 2018 or the leverage of the existing ecosystem to build a solution in a partnership mode (“co-construction”) by Arcep, the French regulator.

On the other hand, for “providers of internet access”, the “remedies” aspect of the EU regulation translates into risks of higher OPEX (fixing issues in limited timeframe), increased subscriber CHURN, and potential cash outlay. Concurrently this creates opportunities for the service providers to deliver the best quality of experience. Hence, the critical need for accurate speed testing and better understanding and management of the Quality of Experience (QoE) for all customers.

Quality of Experience, the right way to measure it on fixed access

When looking for a proxy for Quality of Experience assessment, speed is the usual suspect. Accurate throughput measurement is a must, no doubt. However, real Quality of Experience, especially for fixed access, requires a more holistic approach. Let’s take, xDSL access for example.  There is a tradeoff between the achievable rate and the protection of the line (Signal to Noise Ratio). Low SNR target settings will lead to higher rates but might cause an intermittent connection. The availability of the connection over time or connectivity needs to be considered.

Access networks are by design, oversubscribed. For some networks that share a medium such as GPON or HFC, there is resource contention on the “last mile”. For all networks there is resource contention at the access node (for example, a 1Gbps backhaul link to an OLT (GPON access node) which is servicing 10,000 subscribers.). Therefore, the throughput measurement needs to be automated and scheduled on regular basis to take into account peak hours when contention is the highest. It needs to be compared to advertised speeds and cannot any longer be a theoretical maximum throughput that could be achieved in perfect conditions.

Finally, high latency can impact real-time services and should be closely monitored as well.

Those automated measurements need to eventually be available for all subscribers, on all technologies, on a frequent enough basis and therefore they must be done without interference to customer usage. This means they are required to be affordable and account for cross traffic while not competing with it.

All those factors rule out traditional approaches based on manually triggered tests done only on end user devices  (e.g. APP or web browser) or additional dedicated hardware.

Quality of Experience is, de facto, an end to end concept. However, a specific segment of the network requires special attention: The Home Network and in today’s world home network = Wi-Fi.

The Wi-Fi challenge

Measurements made on large networks show that throughput of in-home Wi-Fi can go very low, making wireless the bottleneck of the whole network. This is particularly true for the 2.4 GHz band as well on the 5 GHz band. The coming of Ultra-fast broadband era unveils Wi-Fi issues.

A few years back, a call to a service provider customer support to complain about the quality would have likely started with this question: “Could you please connect to your router using an Ethernet cable” (just after the classic: “Have you tried turning It off and on again?”). With the explosion in number of mobile devices that connect only in Wi-Fi, service providers need at the very least, visibility on Wi-Fi performance.

When it comes to Wi-Fi performance an obvious indicator is the Received Signal Strength or RSSI.  No one can deny the fact that if you have no signal, the probability of having bad service is high. What is sure is that a good coverage (good RSSI) is not always synonym of good quality. It goes back to the 3 metrics we discussed earlier: Speed/Throughput, Connectivity and Latency. Some might wonder whether the service provider is responsible for the in-home Wi-Fi. Well, the answer is complicated but if service providers take an OPEX motivated and pragmatic approach, the answer is yes. However, service providers need to be able to discriminate between issues arising from their own network or from the in-home Wi-Fi as each are tackled differently.

An ideal solution would provide an end-to-end view by combining the measurement made on the two legs of the network (WAN and Wireless LAN).

Is there a solution on the market fulfilling all those requirements? (spoiler alert: the answer is yes, and this solution can do much more.)

Introducing CloudCheck® TruSpeed

TruSpeed software is designed to enable internet service providers (ISPs) and communications regulatory agencies to measure accurately the available throughput at different network points to ensure the network fulfills end-user expectations, contractual commitments, and advertised speed.  The ASSIA TruSpeed solution leverages ASSIA’s unique methods that allow multi-segment testing. (figure below)

ASSIA’s TruSpeed Solution is built upon 5 components:

  • TruSpeed Agent is a measurement agent installed on the Customer Premise Equipment (modem, residential gateway etc.) that allows scheduled tests to be executed automatically (at peak time and at quiet hours). This agent performs tests that are directed towards test nodes as well as towards single-end testing of the Wi-Fi link.  ASSIA’s TruSpeed software determines whether Wi-Fi is the bottleneck or not.  The TruSpeed agent is currently supported by more than 100 platforms and can virtually be supported by any CPE running Linux.
  • TruSpeed SDK/APP can be used as a measurement agent from the end-user device to a test node for an end-to-end assessment of performance. The end-user can also use the SDK/APP to trigger CPE based measurements. Finally, the APP can provide the end user, through its unique SweetSpots feature, visibility of Wi-Fi throughput and coverage limitations while empowering subscribers to take corrective actions.
  • TruSpeed Server manages the agents, collects all the data sources, and stores them while thereby providing powerful analytics and reports . The server also exposes a set of APIs for easy integration to external tools/OSS/BSS.
  • Access Network Data Collector (ANDC) is an optional module of the TruSpeed Solution. The ANDC allows collection of technical settings provisioned on access nodes (maximum rate, access technology). The ANDC can store the network topology, which can be helpful for correlation and for speed prediction of new customer additions based on neighborhood performance.
  • Speed Test Nodes are smartly placed on or off network for throughput measurement.

CloudCheck TruSpeed is the only solution on the market that can truly address the speed measurement challenges introduced by European Union laws.

For more on TruSpeed introduced by ASSIA today, June 6, 2018, please see our new TruSpeed product page at www.assia-inc.com/products/cloudcheck/truspeed    The press release can be found at www.assia-inc.com/about-us/media-center/


ASSIA’s Approach to Machine Learning/Artificial Intelligence

Tuncay Cil and Haleema Mehmood
Team ASSIA

Posted on May 30, 2018

Part 1: A Primer

Introduction

Arthur Samuel in 1959 defined machine learning as the “field of study that gives computers the ability to learn without being explicitly programmed”. Typically, machine learning and artificial intelligence (AI) are used interchangeably in the industry although AI is a superset concept that is not just limited to machine learning.  Only recently, the advances in cost, processing power and storage on devices and network equipment, have delivered the possibility of AI techniques previously only available for super computer clusters.

In a typical AI/machine learning system for network management, the following steps form the foundations of any prescriptions or predictions that get generated by it:

  • Start with observing a user or a system
  • Describe, categorize and make sense of the observation
  • Attempt to define the behavior of the user or system by establishing a set of baselines
  • Monitor and detect deviations from baselines of defined behaviors
  • Predict future behavior, prescribe trouble-shooting solutions, and self-optimize

Applying these AI techniques to better monitor, detect, troubleshoot and optimize Wi-Fi networks, is part of ASSIA’s core engineering capabilities. While the feasibility of real-world applications of AI techniques is there, the data scientist generalists approach falls short of delivering tangible value to service providers. This is where ASSIA brings real-world performance and value accepted by service providers to the field of AI driven network management.

Basic vocabulary for machine-learning/AI

Data is the input to the overall system and in the context of Wi-Fi management can be anything from physical layer performance metrics, device characteristics and environment, alerts end-user inputs, and Wi-Fi client performance indicators.

A feature is an individual data element relevant to the specific machine learning model and use case. It can be directly extracted from the data or derived after passing the raw data through a pre-processing algorithm. For example, Wi-Fi throughput between an access point and a client can be derived from traffic counters.

A model maps the input features to the desired output or label.  In Wi-Fi management, an output may be the best channel to pick in the user’s environment or the user’s perceived quality of experience score.

Labels augment individual unlabeled data with meaningful tags that are informative. Machine learning models can be applied to the data so that new unlabeled data can be presented to the model and a likely label can be guessed or predicted for that piece of unlabeled data.

Supervised machine learning is a “teaching” technique to develop a relationship between a known set of outputs (labels) and their inputs (features). Once the model is developed, it can be used to predict the output for a new set of inputs.

Training prepares a supervised machine learning model using features extracted from training data along with the appropriate labels. Training can be performed periodically with the update of training data or features.

Unsupervised machine learning is “self-learning,” and no prior training, or preparation is required before it’s deployed. The algorithm automatically discovers relevant structure and relationships among the inputs.

Baselining is a method to determine “expected” behavior for an entity like a user or a system. Once these baselines are established, the models look for deviations from the baselines.

How ASSIA makes machine learning/AI possible

ASSIA’s CloudCheck solution makes AI possible using its agent-cloud architecture. The agent resides on the home Wi-Fi router, collecting the data necessary for meaningful AI. The data is processed and stored in the cloud, allowing for powerful machine learning/AI applications.

For many years, three main issues hindered proper data collection necessary for meaningful cloud-based AI applications in last-mile and Wi-Fi network management:

  • Non-exhaustive platform coverage:There are a variety of CPEs with various chipsets in use in service providers’ networks. It is important to collect data from all types of nodes for network level machine learning. ASSIA solves this problem with its hardware-agnostic CloudCheck agent that provides similar data collection across more than 100 CPE models.
  • Insufficient sampling across time, and network nodes:Overnight data collection that is prevalent in the industry is insufficient for identifying problems in network nodes. Furthermore, it is gravely inadequate for machine learning applications. CloudCheck solves this problem by providing a much higher frequency of data collection. This data is stored and transported efficiently so as to minimize network load.
  • Data collection methods that don’t scale:While trying to collect more data, more frequently, in a more intelligent way, we can’t (yet) assume unlimited computational resources. CloudCheck uses a smart way of mixing event and time triggered data collections that takes into account the dynamic nature of the observed trouble patterns. This resolves the scalability problem of the growing data collection needs without missing critical information.

ASSIA has developed patented data collection techniques that enable application of AI-based algorithms in the real-world environment.

ASSIA’s CloudCheck solution makes use of this data in unsupervised ways to learn baseline behaviors along various dimensions. The learnt policies or thresholds are either fed back to the CloudCheck agent, so it can report deviations from the thresholds to trigger an optimization action, or the policies are used by server algorithms to generate diagnostic outputs. The diagnostic outputs are further used in a supervised learning model that assesses customer satisfaction.

In the next article we will explore specific applications of how ASSIA applies AI and machine learning in Wi-Fi Management. 


What can CloudCheck® solve for the operators?

More than just offering management tools and optimization knobs, CloudCheck helps operators define and address their customer challenges. CloudCheck follows a use-case driven approach, and is purpose built for addressing subscriber issues. Some of the critical operator use-cases are highlighted in bold throughout this article.   CloudCheck does this while driving customer satifcation up and operator costs down.

Deeper visibility into the home for improved customer care.

The proliferation of Wi-Fi devices has extended the last-mile of Internet connectivity further into the home. Operators are now responsible, not only for the broadband service, but also for the user experience on Wi-Fi connected devices. Unfortunately though, operators have little to no visibility into the home networks. CloudCheck provides this visibility by monitoring the Wi-Fi, broadband, and hardware health of the home network and providing diagnostics at various levels of aggregation.

Using machine learning techniques, CloudCheck defines the customer experience in a single Quality of Experience (QoE) Score to help simplify the troubleshooting process and improve customer care efficiency. CloudCheck’s QoE models are trained with call/churn/survey/dispatch data to predict customer behavior and facilitate proactive servicing. For example, consistently low QoE scores can predict customer calls, which allows operators to proactively upgrade those customers to a higher service tier or newer hardware to improve customer satisfaction. Comparing the QoE scores of different groups of customers (e.g. those using router-X vs. those using router-Y) can help the operator with business decisions like vendor selection or new service roll out. Categorical scoring (e.g. Wi-Fi QoE, Broadband QoE, hardware QoE) help identify the category of a problem faced by a customer, and the appropriate fix.  This prevents lengthy service calls and unnecessary technician dispatches.

Improving the wireless performance at home for reducing customer complaints and subscriber churn.

CloudCheck improves customer experience by proactively optimizing the wireless performance of the home network. The band and AP steering solutions work seamlessly over multi-vendor extender/mesh deployments at home to deliver intelligent Wi-Fi roaming for better wireless coverage. With prediction-based channel switching and algorithms to utilize the less-used 5GHz spectrum (DFS channels), CloudCheck improves Wi-Fi speeds and helps in delivering high-speed services (e.g. IPTV) over Wi-Fi.

Broadband services are now being delivered over a mix of technologies, from fiber to DSL to Ethernet to Wi-Fi, and identifying performance bottlenecks (e.g. in the Wi-Fi environment or DSL infrastructure) helps to isolate the problem and apply the proper optimization actions. Hardware stability impacts network performance and identifying faulty hardware using CloudCheck’s system health diagnostics also allows for proactive system and software reboots.

Managing the new challenges from mobile data offloading.

Operators are deploying Wi-Fi hotspots to address the increasing demand for mobile data traffic. Opening home Wi-Fi networks for mobile guest users can congest the wireless network for private in-home users. CloudCheck provides airtime fairness tools that can guarantee QoS for private users by enforcing Wi-Fi level service restrictions on guest connections. As guest network profiles are saved in the mobile phones, private users could connect to the guest network while inside their home. CloudCheck detects such situations and ensures private users are always steered to their private network. CloudCheck can also force low performing guests to roam to a better hotspot or LTE by blocking and/or disconnecting them from the existing guest network.

Enabling self-service and user preferences through a subscriber app.

CloudCheck App’s self-diagnosis features improve user satisfaction by enabling self-help and reducing subscriber dependency on customer service. The popular Wi-Fi SweetSpots feature help’s identify coverage holes in the home and find the appropriate locations to place Wi-Fi access points. The CloudCheck subscriber app also serves as a portal for users to define their service requirements (e.g. monitoring and alerting every disconnection of a Wi-Fi security camera), which can then be merged with the diagnostics policies in the cloud/router, thereby customizing the management policies based on user preferences. As mobile broadband (LTE, 5G) becomes faster, Wi-Fi offloading may not be the default choice due to performance needs. CloudCheck can merge app and router diagnostics to decide whether a mobile device should remain in the LTE environment or switch to a Wi-Fi hotspot.

Adapting to evolving Wi-Fi applications.

CloudCheck can provide application or client specific services based on evolving needs such as voice/video delivery over Wi-Fi and connectivity for IoT devices. CloudCheck can provide targeted optimization for voice/video services such as allocating more airtime for wireless STBs upon detection of video streaming through those special devices. CloudCheck can free up the 2.4GHz band for low-cost IoT devices without dual-band capabilities by steering capable devices to the 5GHz band.

Operators benefit from solving their subscriber problems. With that in mind, CloudCheck looks at call drivers and customer complaints to define critical use-cases and then build the necessary knobs to address them.

Click here to learn more about CloudCheck


Wi-Fi Mesh Extenders: Truth or Hype?

Kevin Mukai
Director Product Marketing, Wi-Fi Solutions

Posted on May 16, 2018

Introduction

ASSIA recently announced support for the Wi-Fi Alliance EasyMesh™ certification program.  Wi-Fi EasyMesh™ delivers a standards-based approach to deploying multiple access points from different vendors, extending uniform Wi-Fi coverage and enhancing performance throughout a larger service area than is possible with a single access point.

Why is ASSIA endorsing Wi-Fi EasyMesh?  First of all, interoperability is good for the service provider market.  A standards-based approach means service providers can achieve more flexibility and reduce costs, while delivering a consistent Wi-Fi management experience.

The mesh and extender product category has been around for some time, which begs the question why another standard is needed for this device category.  I’d like to provide some perspective around this topic by introducing the Top 3 Myths About Mesh and Extenders

 

Myth #1.  Mesh and Extenders are Equivalent

We often interchangeably refer to repeaters and extenders as devices which will boost, repeat, or extend the Wi-Fi signals to improve coverage in the home.   Residential consumers will often place multiple extenders/repeaters in the home to improve coverage.

However, the broad presence of consumer mesh devices in the market has led to some confusion as to what truly is a mesh network versus a multi-access point or extender based network.

A mesh network is often incorrectly defined as merely a configuration consisting of two or more extenders.   Mesh was originally developed as military-based technology to provide wireless coverage across a wide area.  The IEEE defined a mesh technology called 802.11s, which defined packet routing protocols for forwarding and redirecting packets if wireless nodes were to become inactive. While this technology standard has existed, the industry lacked a true interoperability standard for such devices, and hence implementations become proprietary.

In most  residential deployments, mesh nodes typically get configured in a point-to-point or star topology.   So, in reality, “mesh” deployments are typically not even configured as a true mesh network.  Why should a service provider pay for mesh technology if mesh is not even utilized in a deployment?

Instead, service providers can provide a solution that offers maximum speed/capacity of the backhaul to gateway as well as to stations.   Wi-Fi performance is only as good as the weakest link to the end-user device.  Optimizing performance of the fronthaul and backhaul is also essential for maximizing the gains in an extender environment.

 

Myth #2. Mesh Hardware Eliminates the Need for Wi-Fi Management

We have found in our residential deployments that adding extenders often results in no improvement, and in some cases even results in a loss of coverage in the home.   This often occurs for several reasons.  First of all, extenders are often configured improperly and placed in suboptimal locations in the home.  For example, extenders are often placed where the dead coverage spot in a room is actually located, lacking an adequate backhaul connection to the main gateway or AP.  Extenders can also perform poorly because wireless mesh/extenders can lead to greater congestion in the home due to presence of stronger Wi-Fi signals (RSSI) combined with overlapping channels and bands leading to more interference.  Providing proper Wi-Fi management software tools for extender configuration and placement are essential to maximize the benefits of mesh and extenders.

 

Myth #3. Mesh Extenders are the Panacea for Improving  Wi-Fi Performance

Despite the advances in Wi-Fi technology and standards, achieving reliable and adequate performance throughout the home remains a challenge today.  Service providers lack the diagnostics and visibility into the home and often simply resort to deploying mesh/extenders throughout the house to address performance issues.

However, deploying extenders is effective only if the primary issue affecting performance is due to poor signal strength coverage.  Other factors can impact performance such as interference or noise in the environment.   Simply installing  mesh extenders to address non-coverage related performance problems will only exacerbate the subscriber’s Wi-Fi issues.  Accurate diagnostics and determination of the root causes of performance issues is critical for determining if and when a mesh extender should be deployed.

The Wi-Fi industry has come a long way with interoperability standards.  Mesh and extenders can addresses coverage problems in the home when deployed and managed properly.  With Wi-Fi Alliance EasyMesh™, operators can benefit from standards-based hardware solutions combined with the power of ASSIA’s CloudCheck solution  for network wide residential Wi-Fi management.