top of page

Signals through Noise — overcoming Telephony challenges for Connected Car implementation

This multi-part series on examining the challenges of implementing connected vehicle services in India, the issue of cellular network challenges warrants a dedicated piece for itself. The cellular network is the critical component of the connected car ecosystem and in the Indian context, poses some unique challenges for a system designer. Cellular network side constraints force the system designer to make fundamental trade-offs.


Constraints posed by the Telecom Network Layer



The classic 5G triangle has been modified to show how when choosing between having massive number of active devices, providing larger bandwidth for rich data and ensuring very low response times — choosing one or two attributes has an impact on the ability to offer customers the third. Beyond the constraints driven choices, we can look at the impact of the telephony layer on connected vehicles services across further three dimensions:

A. Capacity: Offered vs Available

While MVNOs claim and demonstrate heat maps on GSM availability by band by geographic region, what that ‘availability’ actually means on the ground, especially when further quality of service metrics are added, comes to naught. Quality service deteriorates in a mobility context, due to two vehicle level contexts: a rapidly moving vehicle, when continually transmitting data, will have to switch between network infrastructure; for certain protocols, network providers offer only dynamic IP allocation instead of static IPs. This has implications to a connected vehicles system in terms of session open & closing overheads. In addition to capacity availability, there is no clarity on GSM band fallback options (4G → 3G →2G) available for M2M communication.

B. Cost of Data

Data overheads are the biggest (but often overlooked) input factor in designing connected vehicle systems. M2M data plans are often not as flexible as consumer data plans — a system designer must take adequate considerations to design protocols to ensure spill over of data over agreed plans. MVNOs also have very little incentive to offer better data plans for M2M scenarios due to lowers ARPUs compared to voice based consumer plans.

C. User Expectations

An oft-recurring and avoidable pain point when offering connected vehicles services is expectations of the end users in the Indian market from telematics. We briefly touched upon in our previous article the impact of absence of data as the customer expects: as far as track-and-trace goes, the user expects “real time” updates of their vehicle location; delays in updating vehicle & driver metrics at the end user’s side leads to misleading customer complaints (“tumhara telematics nalayak hain”)


Designing the ecosystem for telephony constraints


There is no silver bullet for overcoming the constraints posed by GSM networks; this requires a concerted set of actions by multiple stakeholders within the connected services ecosystem. But there are several actions that an OEM or a Tier-1 system integrator can do to mitigate these problems:

  1. Existing connected services platforms can leverage data sciences to use the key metrics that make up quality of service, map the outcomes on a geo-spatial layout to determine how network providers really operate. This analysis can then be close-looped with the MNO to either get them to improve their services, or at least hold them accountable for what was promised. The outcome of such an analysis could then be better agreements with MNOs with specific SLAs on metrics like session timeouts, data packet loss etc.

  2. For operating in countries like ours, choice of GSM chipset module becomes very critical for designing a telematics system. It is essential to select a GSM module which has GSM band fallback mechanisms (4G →3G →2G) and negotiate such fallback agreements with the MNO. Specifically for contexts where the connected vehicle operates in remote locations (like agricultural equipment, construction equipment etc.) we recommend even having USSD (Unstructured Supplementary Service Data) as a further fallback for GPRS. Virtually every GSM module being manufactured has AT commands for sending and receiving USSD messages. USSD offers higher QoS levels when compared to SMS, while having greater coverage than GPRS, making it an ideal last resort fallback mechanism for transmitting basic vehicle data.

  3. It is also essential to operate more intelligently with MNOs. For starters, data plans must match the end users’ needs. Users who focus more on vehicle and asset usage reporting and lesser on live location tracking need a lesser transmission frequency and hence may operate on a cheaper data plan. OEMs and Tier-1 SIs must work in a partnership model with MNOs so that their ARPUs get better in the M2M context — value additions like streaming content and digital radio for personal vehicles, add-on services like weather, real time traffic updates and market place data for commercial vehicles make for better end user services, and for better rate plans from the mobile network operator’s perspective.

  4. There are some critical design interventions that need to be made at the telematics device end. The telematics device must be capable of different transmission rates and data structures at different use cases (rapid transmission of minimal data during vehicle running; expanded data sets during diagnostics/alerts etc.). This is where the communication protocol comes in very handy. The device can also do some basic data aggregation and computation of raw vehicle data, along with device side packet compression — this results in significant payload size. Choice of the communication protocol also becomes critical: protocols like UDP are lightweight in terms of session overheads. Using hex files (with adequate security algorithms) will also result in reduced data transmission. There are multiple methods at the device end to handle situations where network availability is poor: from using store-and-forward mechanisms to automatic payload reduction based on signal strength.

Factoring in the amount of investments sunk into existing 2G, 3G and 4G networks, and the availability of silicon from big players like Qualcomm, it is a fair industry consensus that the ubiquitous availability of 5G GSM networks is a good five years away. Until then, the design level interventions discussed in this article are essential to successfully deliver value to end users through connected vehicle services.


This article was originally published in my earlier blog.


bottom of page