Keeping mobile networks up and running is a data intensive job.

Mobile devices and

mobile networks create a lot of data in the course of normal operation. Figure 1
shows just how much demographic, signaling, location, transaction, online and
calling behavior data that is generated daily by 100 million users. Within this
data lies answers to network health issues that impact customer experiences and
break service level agreements. Finding the geographic location of network troubles
is one of the challenges. Knowing the exact base station and even radio at that
base station allows network trouble shooters to quickly resolve radio access
networks (RAN)-related issues. Groundhog Technologies’ CovMo leverages geolocation
data and analytics to pinpoint network health issues for faster resolution. Today,
leading operators with more than one billion subscribers rely on CovMo for the most
efficient resolution of network issues. The growth in 5G standalone (SA) network
deployments results in higher network density, more mobility and high data rates
that translate into larger volumes of data for CovMo to process with tighter real-
time constraints. Groundhog, an Intel® Network Builder Community member, has worked
with Intel to leverage In-Memory Analytics that, combined with the use of
ClickHouse database management system (DBMS), improves data ingest and query
processes boosting CovMo’s ability to process and handle ever larger data sets.
Groundhog Technologies’ CovMo relies on Intel® In-Memory Analytics Acceleration and
4th Gen Intel® Xeon® Scalable CPUs to ensure data ingestion is not a bottleneck for
superior 5G customer experience Mobile Network Operators / Communications Service
Providers Groundhog CovMo Uses Intel® In-Memory Analytics Acceleration for 5G
Solution Brief Figure 1. Daily mobile network data generated by 100 million users.
1Solution Brief | Groundhog CovMo* Uses Intel® In-Memory Analytics Acceleration
for 5G Figure 2. This figure shows CovMo in action, ingesting data from the data
inputs into the mobility intelligence platform, which analyzes the data in order to
inform a variety of applications. CovMo Delivers Mobility Intelligence CovMo is a
cloud-native mobility intelligence solution that provides high-accuracy location
data and analytics for 5G customer experience management and network optimization
(see Figure 2). CovMo can be used to pinpoint issues in a network ranging from the
location of user equipment at all times and the mapping of radio conditions that
could be causing poor user experience. The software is cloud-native with a
microservices architecture that enables a resilient and scalable platform for on-
demand mobile data analytics. CovMo has evolved to support the complexities of 5G
network deployments. CovMo works equally well for indoor and outdoor applications,
which is critical given that around 80% of wireless use is indoors and being
indoors makes geo-localization more difficult. This carrier-grade platform can
process all of the daily network event and user experience data and convert that
into ubiquitous intelligence through its analytics capabilities. One operational
mode is in Open RAN deployments where it is used as an application for a RAN
intelligent controller (RIC) which facilitates third-party applications that
improve network operations. Two RIC types have been defined: near-real time RIC
(near-RT RIC) and the non-real time RIC (non-RT RIC). When deployed as an rAPP in a
non-RT RIC, CovMo can provide services that don’t require real time granularity,
for example radio optimization. When deployed as an xAPP, CovMo provides the full
benefits of near real time data granularity including location intelligence. Some
of the 5G applications for CovMo include: Network Planning and Design: 5G networks
need visibility into network conditions to provide guaranteed service availability.
CovMo data can also help operators detect traffic hotspots that may cause areas of
poor subscriber experience. This data can be used to determine the areas where a 5G
built, CovMo can help to maintain SLAs including high service availability.
Implementation and Troubleshooting: With its near real-time health site reports,
CovMo can be used to validate 5G network design implementations. For example, areas
where interference is caused by a dominant cell, CovMo can facilitate accurate
rectification of the interference. This will grow more important as 5G networks
grow more densely deployed. Monitoring and Optimization: To ensure a quality user
experience, MNOs must have a clear and immediate understanding of end-to-end
network performance. Operators can leverage CovMo’s geolocated network KPI feature
to analyze and fine tune end-to-end 5G network interworking with legacy networks
boosting 5G performance and superior customer experience. Ensuring Data Ingestion
is Not a Bottleneck To help the system ingest terabytes of data, CovMo is
integrated with the open source ClickHouse* a database management system (DBMS)
for real-time online analytical processing (OLAP) applications. ClickHouse is a
columnoriented DBMS which makes it better suited for OLAP applications because it
is faster and uses less compute and storage resources. ClickHouse can quickly scan
large amounts of data which is important for mobile use cases. CovMo Also Leverages
Intel® In-Memory Analytics CovMo also makes use of Intel® In-Memory Analytics, a
feature on all generations of Intel® Xeon® Scalable processors. With Intel In-
Memory Analytics computing architecture, data resides in system memory, so it is
close to the CPU, which reduces latency caused by the time the CPU needs to seek
data on a hard disk and shuttle it close to the CPU. This makes the overall system
much faster. Intel In-Memory Analytics stores data in columns – like ClickHouse
does – for the same performance benefits. Inmemory analytics also supports
massively parallel processing to leverage multi-core and multi-thread processors to
reduce 2 upgrade can have the most impact. Once the 5G network is the latency of
their [Link] Brief | Groundhog CovMo* Uses Intel® In-Memory Analytics
Acceleration for 5G But In-Memory Analytics takes extra CPU cycles away from other
workloads. To free up the CPU and speed up throughput, Intel has built an In-Memory
Analytics Accelerator (IAA) into its 4th Gen Intel Xeon Scalable processor. The IAA
combined with 4th Gen Intel Xeon Scalable processor1 increased ClickHouse
performance2 by 1.59x and reduced memory bandwidth per query by up to 25%. With the
5th Gen Intel ®Xeon® Scalable processor3, database performance increases by 2.49
times. All of the performance results were compared to performance of servers using
3rd Gen Intel® Xeon® Scalable processors. The IAA processes analytic primitives
including scan and filter, cyclic redundancy checks (CRC), compression and
decompression. As a separate on-chip accelerator, the IAA provides more query
throughput while reducing the CPU resources needed. This is especially true for
scan and filter operations that require significant computational performance. The
IAA also provides deeper data compression than software compression which reduces
the bandwidth needed to transport the data to storage and the disk space needed
once it is stored. CovMo also relies on the CPU power of the 4th Gen Intel Xeon
processor for processing complex algorithms using deep learning AI. The processors
provide faster matching and bidding decision time in-between user location, user
interests and the right marketing actions for higher customer satisfaction. CovMo
for 5G Network Build Outs As 5G networks grow in size and density, MNOs can
anticipate QoE challenges that can be addressed by CovMo. CovMo can monitor the
network’s performance against a set of 5G-specific KPIs and analyze for inter-radio
access technology (RAT) to help planners ensure the smooth growth of 5G. CovMo also
improves visibility into network performance allowing MNOs to provide connectivity
for smart healthcare, smart factory, connected frontline, cloud gaming, and smart
drone applications. Using an implementation of CovMo 5G Geolocation supporting both
NSA and SA deployment, MNOs are able to monitor the quality of experience (QoE)
across network generations simultaneously using one centralized platform.
Conclusion 5G networks create a large amount of data that can be used to keep the
network optimized and to detect areas where radio issues are impacting customers.
CovMo provides geolocationbased mobility intelligence platform that can identify
where problems are located for fast resolution. MNOs use CovMo for network
performance but also for spectrum planning and to ensure KPIs are being met. To
ensure CovMo has the compute platform that can keep up with growing 5G networks and
users, it uses ClickHouse to ingest and query large volumes of data. By adding
Intel IAA, built into 4th Gen Intel Xeon Scalable Processors, CovMo can access
unprecedented level of computing to ingest, transform and analyze data at 5G SA
scale.