Understanding / Designing Embedded Systems

Having spent over 18 years developing embedded software for wireless systems, I believe I have developed a certain level of intuition which enabled me to understand the design and problems pretty quickly. This is an attempt to codifying that intuition. I hope to keep this entry live for a few months, will update the last updated date. 

The definition and scope of embedded systems are changing rapidly. We can dilute the definition and settle for something which is "embedded" in a larger system, delivers its functions as per certain rules, has a limited set of specialized functions, have tighter constraints than a general-purpose system, and by extension real-time.  Even a general-purpose system may be viewed as a combination of several embedded systems. Take, for instance, a mobile - where it is getting increasingly difficult to distinguish it from say, a laptop - the cellular modem inside the phone can be very much real-time, a limited set of specialized functions and have extremely tight resource constraints. 

First steps in order to understand an embedded system would be to know

• The processing elements
• Memories
• Power Architecture
• Interfaces

Processing Elements

The Processing elements could be both to execute the SW developed and the processing that happens primarily in hardware blocks. We have the ARM Cores, x86 Cores, Micro Controllers, DSPs, FPGAs where targeted software gets executed. We also have various HW blocks that over a period of time take over the tasks that were earlier executed by SW.  Two major motivations to move tasks to HW are speed, and power consumption - enabler is the confidence to move a certain piece of function to HW. For example, most high-end phones today implement Camera processing / JPEG encoding etc in HW.  Similarly, there is a lot of custom HW to accelerate neural network inferences. 

Clearly, when we look at processing elements, we need to understand the HW/SW split and how the hardware is programmed, and by whom. Also of interest would be the speed at which each of these processing elements operates and why there are so many processing elements that can do the same task. For example - in a mobile or a laptop, a CPU Core (ARM/x86) can do whatever task a GPU can do. Yet, GPUs are ubiquitous and we know why. Likewise, early smartphones used GPUs for AI tasks - but as the asks increased we moved to DSP-based accelerators and AI accelerators. An easy way to understand would be to take a Product Requirement Document (PRD) and understand which processing element executes each of the product requirements. 

Memories

A typical system can have access to several types of memories. On-chip memories such as Level1,2,3  and System caches, Non-Volatile Memory (which is retained on reboots), Read Only Memory (ROM), RAM, Secondary Storage, EFPROM for fuses, and so on. 

We need to understand the rationale for each of the memory's availability and sizing. A lot of work goes during the system design phase to understand if a processing block needs a particular amount of allocation in L1, L2, L3 and System Caches. The latency increases as you move across the memory hierarchy but the price reduces. The system performance teams, if they have a base chip to compare - can fairly accurately come up with required memories based on the feature delta compared to the base chip. A combination of increasing DDR speeds as we go from LPDDR to LPDDR5 and the efficiency of moving certain functionality from core-specific caches to system caches has caused changes in the sizing of caches in the hierarchy. 

Power Architecture

There was a time when Power was considered to be important only for small battery-operated devices like mobile phones. Increasing device complexity that is pushing the power envelope higher and higher and thereby increasing the costs of cooling solutions is motivating designers to consider optimum power performance even for use cases where traditionally power wasn't a concern. For example, laptops consuming lesser power can go with a fanless design and reduced battery sizing for the same battery backup - both of which have an impact on the weight and cost of the laptop. On the other hand, Data centers are extremely power-guzzling that any saving there could mean a lot of CAPEX saved in cooling solution. In the industrial setting - when the system is in a remote area without access to main power, there is a very strong motivation to make systems run on a set of AA batteries for the life of the system. 

While the overall power consumption as mentioned above is important to understand - it is also important to note which processing elements will be powered and how they will be power gated during idle stages. When a particular processing element is in a power gated state, how do the interrupts targeted to that core get handled, can the interrupts wakeup the target core? How many levels of sleep are possible. What is the deepest level and what is the power consumption during that level could be some questions that can give some understanding of the system design. 

Interfaces

When we think of the interfaces - for mobile phones, the interfaces are what the user can use to communicate - touch and other sensors such as accelerometer etc, speech, input with on-screen keyboard, key presses etc. Typically a complex system such as a mobile phone has multiple chips apart from the main digital chip that takes care of most of the functionality. There are chips to supply regulated power from the battery known as Power Management ICs, there are charger chips to enable charging from the power source, there are external codecs and speakers for better audio performance, there are Radio Frequency chips for cellular, BT, FM, and WLAN technologies, cameras and so on. Within the main silicon as well, there will be typically multiple cores which perform target functions - how each of these cores communicate with the other cores within the silicon on chip (SoC) and also with chips outside the SoC is another area that will provide a lot of perspective of the design. 

Adequate Retirement Fund - Indian Context

Introduction

This is an assessment of the amount of retirement corpus required in India when one wants to stop actively working for money. I don't mean retirement as a state of being completely out of work, it is a state where the monetary rewards are not the primary motives for the activities one takes up. This article heavily borrows the ideas from The Trinity Study and adapts them to the Indian context. The idea is simple but it also allows for several risks to be covered by taking appropriate insurances and allowing a small contingency fund that is outside of monthly withdrawals. 

The Idea

The Trinity Study is based on the analysis that S&P long term returns - Compounded Annual Growth Rate (CAGR) exceeds the inflation by 400 basis points or 4%. If one plans to retire today with a desire to leave a reasonable inheritance, one's corpus needs would be (Current Annual Expenses * 100 / 4). This will ensure that the corpus never depletes to zero even if the monthly withdrawals are inflation-adjusted. 

In order to arrive at the corpus, one needs to know today's annual expenditure and when one wants to retire. The calculation assumes a certain return from equities, a certain inflation percentage, and proposes 7% as an acceptable withdrawal amount in the Indian Context. This puts the corpus required at  (Annual expenses *100 / 7). If you believe that India will move from developing to the developed economy during your retirement, you can either plan to factor 4% or even 3.5% withdrawal rates from the corpus. This simply increases the corpus needs by a factor of 2. 

The Data Used

The model is heavily based on the expected returns on the Corpus and the expected inflation. 

Returns: Nifty returns over the last 19 years (Jan 2001 - Dec 2019) have been used in this computation. This covers two major slowdowns in the stock market one in 2001-02 and the other in 2008-09 and several other short term bear and bull cycles. The choice of this period also excludes the steep rise in stock markets due to the opening up of the economy that happened during the decade before (1990-2000).

Nifty Closing values on 1st Jan 2001 is 1254 and that on 31st Dec 2019 is 12168 from  Nifty Historical Data. This translates to a CAGR of 12.7%. 

Cost Inflation Index (CII) from the Income Tax Department, Government of India has been chosen as the Inflation index. This has been chosen as it has the base year of 2001-02 which nicely aligns with the Nifty period. CPI and WPI may have been more appropriate but they weren't picked due to changes in the base year and the inherent volatility of those indices. CII for 2019-20 is 289, it implies an inflation CAGR of 5.7% over the last 19 years. 

If the withdrawal rate is 7% (12.7%-5.7%)  - the corpus will last an eternity as long as the inflation-adjusted annual spend is constant and over a long period (return - inflation) remains constant. 

Recommended Corpus

[Conversion: 1 lakh = 100,000 & 1 Crore = 10,000,000 i.e., 10 million. All calculations are in Indian Rupees. As per statistics released by the Income Tax Department for 2018-19, less than 0.2 million (0.34%) individuals among the 58.7 million assesses declared income of more than 5 million rupees. The 3 mil & 4 mil Rupees (30 lakhs and 40 lakhs) used below go a long way with respect to purchasing power in India if major mortgages such as house are paid up]

The key input to the calculation of the corpus is the annual expense estimate. IT makes sense to amortize big-ticket replacement costs on an annual basis and include them in the annual expense.

Say you have two cars, costing a total of 50 lakh rupees. You have to replace them after 10 years at a salvage value of 8 lakhs in today's value. You can amortize the 42 lakhs replacement cost over 10 years and add it to your annual expense. That would be around 2.5 lakhs per annum. You can include it in the annual expense and transfer the inflation-adjusted equivalent of 2.5 lakhs each year to a separate fund and withdraw it when it is time to replace. 

All major anticipated expenses such as kids' education, marriages, and capital expenses towards the primary home may be amortized as above and included in the annual expense. Wherever there is an option to cover risks by insuring appropriately such as life, health, and property - it has to be used. A small contingency fund to cover expenses beyond insurance limits and covers needs to be planned. 

Retiring Today

Corpus required for an annual spend of

• 30 lakhs => 4.3 Cr  [calculation: 7% of 4.3 Cr = 30 lakhs]
• 40 lakhs => 5.7 Cr

Snippet from easy-calc on SWP (Systematic Withdrawl Plan) from a Corpus of 4.3 Crores with monthly withdrawal starting at 2.5 lakhs and increasing monthly at the inflation rate is below. 

Retiring in 10 years

The 4.3 Cr and 5.7 Cr need to be adjusted for 5.7% inflation. So one needs a corpus of 7.5 Cr and 9.9 Cr respectively if one plans to retire after 10 years. In case you want to start at zero today to build 7.5 Cr in 10 years, you need a SIP of 3.1 lakhs per month and for 9.9 Cr, it will be 4.1 lakhs per month. 

Snippet from SIP Calculator below

* Please note that the above numbers are for the worst-case scenario, the expectation is that one has some retirement corpus and also the mandatory provident fund accumulations are also significant (25 lakhs - 1.5 Cr) during a 10 year period based on your salary and contributions. Additionally, the activities taken up during retirement may also provide some small monetary gain. 

Retiring in 20 years

The adjusted Corpus values for 30 lakh and 40 lakh annual expenses are 13 Cr & 17.3 Cr [translating to monthly SIPs of 1.18 lakhs and 1.57 lakhs] respectively. It may be noted that the SIP contributions required are lesser due to longer duration. 

Possible Extensions

The above one is simplistic in the sense that it factors only stock market returns for both Systematic Investment and Withdrawal. One can extend the concept to have a comfortable representation of equities, debt, return yielding real estate, and other instruments such as gold/silver bars in both accumulation and consumption phases. 

Several large and mid-cap mutual funds try to and succeed in beating the NIFTY returns. If you factor higher returns, you may need to make lesser monthly contributions to the SIP.

The corpus accumulated in Mutual funds via SIPs may be used to during the withdrawal period by converting into SWP.

Long term capital gains and other taxes are not factored in the above illustration. @10% tax, it is actually 27 lakhs per annum in hand instead of 30 lakhs in the above computation. 

There are several articles over the web that study the continued relevance of the trinity study and improve it to include various risks and even recommend reduced withdrawal rates: link. 

References

1. https://en.wikipedia.org/wiki/Trinity_study
2. https://sipcalculator.in/
3. http://easy-calc.com/financial-calculators/SIP/Advance-SWP-Calculator
4. https://www.incometaxindia.gov.in/Charts%20%20Tables/Cost-Inflation-Index.htm
5. https://www.niftyindices.com/reports/historical-data

Healthy Body and Mind

This is where I want to summarize my thoughts on the good health of body and mind.

Body

There are three dimensions of body health, I read several years ago. Strength, Flexibility, and Endurance. Easy and obvious examples: Strength allows one to lift more weights, Flexibility allows one to touch one's tows without bending knees and Endurance to run for miles. If one is good at all three dimensions he can clearly be seen as having good physical health. I was told an hour of physical activity is needed for all. But we have to qualify it better. Say, gentle strolling for an hour does only burn calories - one can argue that it would improve endurance, but it does nothing for strength and Flexibility. Then comes the workout for focussed muscle groups interspersed with Cardio. This covers Strength and Endurance but leaves out endurance. Likewise playing any racquet sport is more of endurance a bit of strength and not much of flexibility. It gets tougher when you can't even spare an hour. After toying with various forms of workout - I realized Surya Namaskaras (in Yoga) is the one that covers all three. Just 30 mins, non-stop Surya Namaskaras (Sun Salutations) gave me the same effect as 20 mins each of strength, flexibility and endurance exercises.

Mind

Now comes the mind - what are dimensions of mental fitness. Let's start with the definition -According to the WHO (World Health Organization), mental health is: "... a state of well-being in which the individual realizes his or her own abilities, can cope with the normal stresses of life, can work productively and fruitfully, and is able to make a contribution to his or her community."

How does one realize his or her own abilities?  What good is a skill if it is not put to use productively and fruitfully? What good is a life if months and years that have passed by don't give a sense of life well spent. "Life well spent" has personal definitions that can obviously change from person to person but also along the time axis for the same person. Essentially, this looks broadly in line with the self actualization needs in Maslow's hierarchy.


Maslow's Hierarchy of needs [Ref: https://www.simplypsychology.org/maslow.html]

There are several studies happening under Cognitive NeuroScience to figure out the differences in neural activity for sensory responses vs thoughts / moral judgments. One interesting video on the topic by Rebecca Saxe is here: link.

As I think through it, it seems the three dimensions of mental health would be Intellect, Memory and Focus. If you belong to the school of thought that the mind actions (thoughts, moral judgments etc) all occur in the brain - it would be easy to infer that the strong body is fundamental for a solid mind. Meditation has been found to help people with improving both memory and focus.

I hope to continue to work further on this topic to figure out further means to improve mental health.

Reasoning God

Prologue
Just for a little background - I was born in a Hindu family. My parents are very religious but they never forced it upon us - they did introduce it though. In fact my sister was very attracted to Christianity for several years and my parents never had a problem. My father especially is a devout Hindu. I have never seen him miss his daily prayer of an hour in the morning. Much of it is chanting the praises of God. He is not particularly ritualistic barring lighting the lamp and offering Harati (using camphor)  to the God. I have always looked up to my father and so as a kid I used to pray the same way (not for as long, though). That's probably the beginning of my tryst with God.

God over the years
My religious beliefs / monologues with God could be broadly classified into three phases. I mention the dominant reason in each of the phases even though there is a streak of all three all through.

• Up to around 12 years of age: I used to believe that prayer is something through which I can get things from God. I used to pray for good grades at school, for things I wanted and also sometimes good things for the people I liked. I had believed He would oblige when I pray. As a kid it never occurred to me how could the supreme being be flattered.
   
• 12 to 20 years of age: This is the period when I started asking myself why should God oblige even if He exists? What way am I special compared to the several million others who do the same or more sincere prayer, even if I buy into the logic that He treats all of us as his children. I sort of felt that if God is there, he would love his children who are 'good', 'do their bit', not definitely those who just praise Him ...This is also the period when I started trying to reason each of the rituals I / my family followed. Over the years I started reasoning that I needed to trust God to give the cent percent to what I am doing at that moment .... especially when the odds are against. Its only the belief in God that gave me strength to move on in trying times. I have never been able to convince myself that God existed, so I tried to reason that God doesn't exist - I couldn't do that as well - so I left that question. None of these have never stopped me from offering prayers. However,  the purpose of prayer gradually started  drifting from 'to please God'  to ' to keep thinking of God ' and 'thanking God '
• 20 years to present: Over the years I gradually started reasoning that it is the power of strong desire which is the most essential in making people lucky. Obviously the strength of desire is proportional to the belief of the likelihood of it happening - which is easy if the omnipotent is by your side and ready to pitch-in when needed. Recently I read this book "The Secret--Rhonda Byrne" and  found the whole 'attracting positive energy' concept pretty interesting. I still believe in the existence of a supreme power thats maintaining overall sanity in the world we live in but not so much in the shape or stories around It. Thinking that we are in some way related to the Supreme Power, allows a possibility of becoming that super power as well. One hypothetical approach is to get the traits of God - just like what all miracles do we expect to occur before believing when God presents Himself to us. A humanly impossible trait is to  certainly be excellent in each and everything. Then the inference would be to striving for excellence in each and every thing one does should get one closer to becoming the supreme power.
  
  So I started believing that, by continuously bettering myself I can become God. The flaw in the logic is that 'good' is not absolute - it is defined for the existence and continuation of society and hence specific to a particular one -  all the 'accepted' virtues seem to have as the core purpose the continuation of society. Clearly good is not the way to reach God for He is absolute and beyond society, region or race. Then the only way I could get better is by having a control over my mind and body for my thoughts and actions (the 'I ' here can't be the mind or the body - has to be something else and for want of anything better I would think 'I' as the soul here).  I find the Bhagavad Gita thought of eternity of soul strongly appealing.
  
  I still have a lot of loose ends in my thought process - but to consolidate as of this writing (2012)  - The soul is the form of God, it exists in everything that exists .. it may be convenient to assume soul as the piece of God within. All this writing is a result of thinking. Mind is what I believe is thinking and it resides in my brain. Clearly I have not realized the existence of my soul. Its easy to see the inseparability of body and mind if thoughts (the works of mind) are a result of the processed inputs from the senses. I don't know the path to connecting to my soul - the introspection continues ........ but I know when I can connect to my soul (that is when my mind body and soul becomes one) - I am God ....

Epilogue:
As a side note -  I tried to put myself in the shoes of God and see how would He react when faced with a certain situation. Very soon it stopped making sense because God influences/works on the aggregate.
Otherwise, I still pray to God - and it is extremely convenient to do it the religious way I have been used to - though not a voracious reader whenever I get time I have been updating myself of various schools of spirituality and forms of yoga. It is still making sense to visit temples because I feel several of my fleeting questions get convincingly answered during these trips- a heightened thought of God could be a reason. I did try to also reason idol worship, presence of several Gods (and even more, families of Gods)  and some of the rituals that made sense to me - I will write about them some  other time.

The evolution of my thoughts on this subject would not have been possible if not for my discussions with Hima (the other half), Manjusha and a few of my friends and colleagues.

What could Google do with Motorola Mobility

Google has recently acquired Motorola Mobility for US$12.5 billion. It is very clear that Google has acquired primarily for the 24K+ patents the company had in the wireless space. The patents are important for Google to protect the Android adopters from the lawsuits of the likes of Microsoft. In this post I will examine ways in which Google can make the best of the investment in the near term, medium term and long term.

Near term: Atleast for another year till the regulatory approvals are through, the business has to be as usual. Also for all the phones Motorola Mobility has sold they need to support for a minimum of 3 years (typical life of phone). Motorola's smart phones are primarily Android based, so they can continue without any issue. Google should let Motorola continue developing their LTE (3.9G) and 4G solutions. Google can also look at potential acquisitions and agreements for the missing pieces of the phone platform.

Medium term: In the next 2-3 years Google should focus on coming up with an entire platform integrated from the RF (antenna) all the way upto Android OS. Google could make use of their software design capabilities and ensure that any small vendor can quickly make an Android phone with that platform. The platform should have completed the operator acceptance and GCF type approval tests mandatory for launch in several markets. The idea is to essentially accelerate the commoditization of the mobile handset industry. As mentioned in another post in the same blog, Mediatek is doing something similar for GSM. What would be important at this point of time is to reduce the Motorola portfolio of mobile devices to probably something like the nexus series and transfer the brand equity to the platform developed. So the mobiles apart from the vendor name could also sport things like Powered by Motorola. There is however a big risk of competing with the likes of Intel Mobile and Qualcomm. Unless Android gets as big as the current Windows for Desktops and Laptops, it may not be very easy to fight the likes of Qualcomm and Intel.

Long term: In 5+ years, Google could continue investing in the next generation wireless technologies, include the hardware that enables more secure and seem less cloud experience in their platforms and support localization and customization of handsets. Google has to evolve Android such that substantial customization is possible and yet apps developed for one version of Android run across the customizations. By enabling smaller players to enter mobile handset development Google can ensure that they retain control over the Android version that goes into the market, for rarely do small vendors have the where withal for major Android abstraction development. The entry into the platform business will allow Google more flexibility to evolve systems to suite the next generation services (for example the GPS chip on phone can communicate better with the services running on Android).

Thus in my opinion it is important that Google should continue acquisitions in this space if they have to maintain their customer reach.

PS: No strategy can be made in isolation to the environment. Especially in this space where all the big names with deep pockets are operating. Also every action of Google will be observed and analysed thoroughly. For example, Apple may decide to become more open or if Microsoft-Nokia's third ecosystem clicks they can present a different set of challenges all together.

Non Linear Revenue Models - Indian IT Sector

Introduction

One of the biggest threats faced by the Indian IT industry is the linearity in the employee count and revenues. As of 2011 the top five Indian IT outsourcing companies have around 100,000 employees on board (some less and some more). One can see a direct proportion between the revenues and the employees and typically over the years the ratio has been 35,000 US$ to 45,000 US$ per employee per year [source: annual reports of the companies]. Gowing byt he current gowth trends of the top five IT services companies from India, for growth rate to continue each of the companies would need a workforce of more than two million (same as the current strength of Peoples Liberation Army, China the largest Army in the world) by the year 2031 - which is an unlikely solution.

Nonlinear Revenue models (Service Providers Perspective)

There has been a strong drive among the IT service providers especially from India to unlink revenues and the head count. The top companies have been trying several ideas with varying degrees of success. In this section I describe five of the broad changes needed from a Service Providers perspective.

1. New Approaches
   While the two approaches discussed here are not entirely new but not many companies make significant revenues in these models compared to their time tested Time and Material and Fixed Price models.
   
   First, Creation of IP Blocks - There is a possibility of companies to create IP blocks which could be sold to / integrated with the products of several clients. In the telecom sector the IP block can be a simple codec implementation to a complete protocol stack for some of the emerging technologies. Another example could be in the Cloud space - companies can invest in IP blocks for rapidly rolling out the internal clouds. One strategy for the choice of the IP block could be something that the customer sees as good alternate option. For time to market and or for matching the competitor offering reasons clients may be willing to pick the IP block.
   Consider the case of Samsung - it was a part of both Open Handset Alliance (Android) and Symbian Foundation as well while having its own BaDa Operating System. There are several hardware platforms in the market each of which takes several months of effort to integrate with a particular mobile OS. The platform vendors also can not integrate their solution with each of the operating sytems. While Samsung as a part of its strategy may invest in several competing Operating systems- keep the one that does well and dump everything else that doesn't, not every company has the business strategy or deep pockets of Samsung. Services companies can partner with some of the phone platform vendors and integrate the platform with some of the Operating systems. If the investment pays off the handset OEM will be willing to agree for a royalty based payment structure, if not the services company can always showcase it as a part of their capability demonstration in bidding for the usual T&M / Fixed Price projects.
   
   Second, Creation of Platforms - A platform is by itself is unsaleable and needs associated customization to meet the needs of clients. Large software companies are in a better position to create platforms. One of the possible approaches could be to work with a client to develop a solution on top of a platform and negotiate the pricing such that the Services firm keeps the rights of the platform and the first client keeps the rights of the solution built on top. Client may be preferentially billed for the support provided in IP creation. It is however very important to invest in the right technology / project to be chosen for something like this. Typical examples could mean features not really creating competitive advantage but are mandatory for running the business such as billing, policy compliance etc (in case of telecom companies). The clearly fragmented OSS / BSS market makes it a potential case for this. The availability of a platform for rapid application development will reduce the switching costs for network operators - thus keeps them in a better bargaining position with the OSS/BSS vendors. As a matter of fact there are already several Cloud offerings in this space. For the app stores, the Infosys Flypp story is another example for good platform idea.
   
   
2. New Markets:
   Services companies still have their structures, business models and policies aligned to doing business with North American and European corporations and governments. There are several other markets especially the emerging one which may need a completely different set of business model for them to succeed. Services companies choosing to invest in the new markets have a native advantage as the market grows. One example again in Telecom could be the switches of less than 100,000 lines capacities C-Dot (a government company) exports to several emerging economies. And the reselling old GSM network equipment to emerging economies when countries adopted 3G is another example. While most of these cases may look like potentially less profitable avenues - they can be easily converted into royalty based revenue streams by enabling smaller local businesses to work on these in their cost-structures while providing only the technical / managerial capability. FabIndia's business model may have takers for instance.
   
   
3. Pricing Models:
   Apart from the pricing aspects of Fixed Price and Time & Material models companies may consider the following models as well.
   Business Outcome based model - For projects in which the services companies have strong experience and cases where the companies have caused clear business benefits in the past make a case for this. A portion of the benefit provided can be claimed as the fees. Companies should be careful while choosing such projects. Projects causing head count reduction / highly transformational projects with significant people interaction of employees of the client companies may be avoided. Projects that simply convert capital and fixed costs into variable costs such as implementation of Cloud etc could be potential example candidates. Such projects when coupled with the IP blocks and Platforms described in the earlier sections make a cohesive strategy. Highly specialised services in the domain of business intelligence helping companies plan their production, supply chain and pricing could also be another class of example. While there are consulting companies to advice - there are n't many to take a share of both Profits and Losses and at the same time follow the execution.
   
   Taking complete / partial ownership of Client Product / Service lines: Companies exiting businesses have an obligation of support for the agreed product lines. There have also been examples where larger companies lend their brand to the right offering. While white lable manufacturing is not considered very lucrative, picking up the right project inline with the competencies and existing IP may give rise to a non-linear revenue possibility. In fact, companies can specialise in taking up complete / partial ownerships for products in mature or decline stages of the lifecycle. Companies can use their market reach if they are present in markets other than the client's strong / choice markets. Again caution should be exercised in the choice of the product and the market so that the company can leverage its strengths for those incremental revenues. One example in the media space could be partnering with clients operating in the internet space unwilling to cater to specific markets, if the market segment could be profitably served by the company.
   
   
   
4. Incentives and Measurements:
   Stock Market expectations and Senior management incentive structures linked to them are considered one of the strongest reasons for risk averse nature of companies. I was told that there are sales teams that prefer selling a client's product over in-house product due to incentive structures.
   Measurement based on quarterly or even annual revenues and profits may not work well if investments have to happen in the non-linear revenue generating businesses. It is certainly not an easy task to device proper incentive structures but is a very important aspect. Certain exponential returns may be considered on business success which ensure that there is appropriate reward for the risk taken. It has been experienced by companies that it is difficult to attract talent across the experience levels for risky ventures. OB researchers may attribute it to the lack of social security in the Indian context. Creating subsidiaries with P&L responsibilities may be considered after the incubation period. The companies may need to start operations outside India if the availability of talent with the required risk appetite is available. Here the intention is not to typecast Indian work force as risk averse but to see in which geography can results be achieved in the most cost-effective fashion with the least business risk.
   
   
   
   
5. Organisation Structure:
   The Organisation Structures should be aligned to properly take up the non-linear initiatives. There have been debates if the Finacle model is better or the OnMobile model. Irrespective of the model the structure of organisation should try to get the best of both worlds - Firstly, the creation of subsidiaries which are smaller and nimbler allow quick response to the opportunities and convert them into profitable business without the bureaucratic delays unavoidable in the larger parent organisations of today. Two, it would be unfortunate and injustice to the vast amount of knowledge of customers and markets built by the wider organization is not utilized. This brings a challenge that while the smaller subsidiaries have the necessary autonomy, it should also be ensured that they are not operating in silos.

Acknowledgements

Some of the ideas in this blog are result of the discussions I had with faculty, peer group and industry veterans during the CSITM Participatory Research Workshop conducted at IIM Bangalore in May 2011. While there have been several interesting ideas and concepts discussed - this post is dedicated to the Service Provider's perspective (the sub-group in the work shop in which I have contributed my thoughts). Errors, omissions and interpretations of examples from the Telecom Sector may be entirely attributed to the author.

The mobile handset industry

INTRODUCTION

In this blog I will first attempt to explain the mobile handset industry as I see it in 2011. After some serious thinking of creating my own framework, I felt probably keeping it simple by using the very popular Porter's five forces frame work taught in my first semester MBA course makes more sense, afterall Mr. Porter is not popular without a reason.

In the end I will attempt to give some possible future directions of the industry and certain strategic choices the players have. I will use examples from the Indian handset industry but I think in this globalised world it can be applied to several similar markets (more precisely similar distribution channel structures primarily dominated by retail and not much of operator subsidy).

I have been reading a lot about the mobile industry (I can call myself an industry insider with 8.5 years of experience developing software and systems across handset, chipset and network infrastructure companies ) so some of my statements which are very obvious to me may not be so much to outsiders. One aspect MBA teaches anybody is to always have supporting data - I have given credit to the data sources towards the end of the post.

PORTER'S FIVE FORCES

So here are the five forces that I am using to evaluate the attractiveness of an industry

(1) Threat of Entry

(2) Bargaining Power of Buyers

(3) Bargaining Power of Suppliers

(4) Threat of Substitutes

(5) Intensity of Rivalry

(1) Threat of Entry

This is one area which has seen a dramatic change just over the last three years. When I along with a couple of my classmates worked on the same topic in 2008 - we termed the Threat of Entry to be low. At that time there were only these huge companies Nokia, Samsung, Motorola, Sony Ericsson, RIM (Blackberry) and Apple. Then there were huge capital requirements, economies of scale advantages in production, distribution and after sale service, brand and technical knowhow to integrate several hardware and software bits which prohibited several players from entering the market.

The reduction of entry barriers is evident in the Global handset market share. From the Strategy Analytics data on the Market shares one can see that the share of other handset vendors after the top 8 globally has raised from 16.3% in 2009 to 22.7% in 2010. The share of other handset vendors is highest in APAC (Asia-Pacific) at 31.2% in 2009 and grown to 36.5% in 2010. Latin America, Africa and Middle East regions also have more than 20% other vendor shares in 2010 whereas North America, Europe have less than 10%. One of the possible reasons is the relative weight of distribution channels in these markets[2].

Cases from India (Micromax & Karbonn):

Micromax, which has started its mobile operations in 2008 is close to the second largest handset vendor in India Samsung. Started around the same time Karbonn sold an average 60o,000 phones a month in 2010. The figures are not very far compared to the market leader Nokia's approximate 4 million phone a month. From the Draft Red Herring Prospectus of Micromax [3] - one can see that while the comany's revenues grew from 375 Crore Rupees (83 mil US$) to 1662 crore Rupees(369 mil US$) an astounding 345% growth the net profits are at around 12% of revenues for 2010. The company depends for every thing including the hardware, software and service on third party vendors.

[Side note: Micromax is a great success story. Though the PAT is not great, on their meager asset base and share capital they have delivered an amazing RoE. The ease of entry and lesser profitability do not give a feel that even 20-25% growth rates are sustainable]

One of the companies enabling this entry of new OEMs is Mediatek. Mediatek has about 35% market share in the baseband chips used in the GSM phones. Over the years Mediatek has developed a set of baseband SoC (system on Chip) solutions with internal R&D and by acquisitions (Cellular Divison of Andalog Devices) and licensing agreements (Protocol stack of Sasken) Significant portion of MediaTek's work force support the smaller OEMs with software and integration services allowing them to cut down the time to market and the investment needed.

The role of brand and distribution network as key success factors has reduced significantly.

To summarise one can say that the Threat of Entry is high.

(2) Bargaining Power of Buyers

Globally it may be visualised as three broad categories of buyer groups with differing levels of influence

- Network operators - Significant share of buying especially in North America and Europe happens through the Network operators. In most of the cases the network operators have good bargaining power over the handset manufacturers.

- Large electronic stores - Large chain stores exert medium bargaining power over the handset vendors due to the limited shelf space and the number of models of handsets available.

- End users- End users though individually have relatively less bargaining power, the availability of choice and the overall drive towards commoditization allow the customer get more value for the price.

In summary it may be said that the Bargaining Power of Buyers is Medium.

(3) Bargaining Power of Suppliers

The handset / tablet parts can be seen as 5 major cost components

1) Baseband, application processor,RF and amplifiers

2) Connectivity solutions like Wireless Lan(Wi-Fi), Bluetooth, GPS, USB, HDMI, DLNA etc.

3) Display

4) Memory

5) Operating System

1) Baseband, application processor etc

The market leader (Qualcomm) has just 20-25% market share in volume. This may make one mistake that the bargaining power of Qualcomm may be less. But Qualcomm also has a revenue share in upwards of 40-45%. Even after discounting the IP royalties from their CDMA technology, the revenue share is still high. The close second MediaTek has 35% share in all GSM/GPRS/EGPRS shipments. Even though the revenue share of Mediatek is less , as a company Mediatek has been posting net profits around 20% every year. So it can be said that the bargaining power is with the suppliers in some cases and with the OEMs in some other cases.

2) Connectivity solutions

The solutions are already extremely commoditized and the end user is certainly not aware of which one is the supplier of the GPS, Bluetooth, WiFi components. There were times in the past when CSR (Cambridge Silicon Radio) used to have a near monopoly in Bluetooth solution (about 60% market share) but it no longer is valid.

So here we may say that the bargaining power of the supplier is less.

3) Display

Here again some display companies and technologies such as AMOLED have demand outstripping supply. So in high end smart phones where AMOLED finds its use majorly the bargaining power is with the suppliers. Otherwise it is with the OEMs.

4) Memory

Due to heavy competition in the Memory industry and crashing of prices by the Korean manufacturers one may say that as of now the bargaining power lies with the OEMs in this space.

5) Operating Systems

Two of the major operating systems (Android and Symbian) are free. While Apple iOS is closed, Microsoft has a very less percentage share. So one may say that purely in terms of cost measures there is no question of bargaining power.

In summary it may be stated that the bargaining power of the suppliers is medium.

(4) Threat of Substitutes

There are no real substitutes if we consider Tablets (iPads, Samsung Galaxy Tab etc) as a subset of Smartphones.

(5) Intensity of Rivalry

The industry has seen new comers competing on price and some new comers competing on differentiation and the incumbents are facing pressure. Especially for instance Nokia which has offering across the price points is hit by new comers (offering sub 100$ phones based on MediaTek chipset) in the low end and by the Apple I-Phone and other Android smart phones in the high end.

Even if one takes the profitability of the OEMs, except Apple no company has profitability around 20% which used to be the case during 2001 - 2005.

There is also this new influx of laptop vendors entering the smart phone arena with their own distribution networks and existing clientele.

SUMMARY

In summary it can be said that the mobile handset industry is loosing its attractiveness over the period. It appears to be entering a periods of commoditization where there are no significant differences between offerings at any price point.

Thanks to the number of discrete chips that go into a mobile phone, there is still a significant quality vs price vs brand equation at play. For example one can get a GPS solution across the price bands and certainly the GPS chip used in a sub 100$ phone takes visibly longer time to latch onto a GPS signal than a 400$ smartphone. Similar things may be said about the receiver performance, appeal of the screen and so on.

However for a given price point there is very little difference between offerings containing the same features. So the value customers willing to pay to a specific brand over other has been reducing.

Derived products by customizing the base product to wider groups, Designing Value Adding Services around the product and offering a solution are some of the methods that have been widely adopted to beat the commoditization spiral. For example - companies making sugar cubes made more profits than the companies selling raw sugar. OEMs now have a compulsion to communicate and deliver value to their customers and thereby differentiate their products. While there have been some efforts in the past (for example Nokia's Supernova series with a mirror finish home screen targeted to women) which were received well by the market. But it is high time OEMs make more customized products for individual customer groups and provide the appropriate value. It does reduce the economies of scale advantages but it is probably need of the hour to differentiate.

REFERENCES

(1) Porters Five Forces - http://en.wikipedia.org/wiki/Porter_five_forces_analysis

[For my love for Wikipedia :) .

The original HBR link - http://hbr.org/1979/03/how-competitive-forces-shape-strategy/ar/1 - if you are in (or alum of) a B-School you may already have access to this article]

(2) Strategy Analytics - http://www.strategyanalytics.com/default.aspx?mod=reportabstractviewer&a0=6451

(3) Micromax - DRHP - http://www.sebi.gov.in/dp/micromaxdraft.pdf

(4) Invasion of Price Warriors - Business Today - 16 May 2010.