Monday, April 25, 2011

Simulating Success

There is a way to simulate apparantly incalculable phenomena such as predicting movements and share prices in the stockmarket or predicting weather changes. Let's say you roll a dice, how do you know which number you will get on a given roll? Using the Monte Carlo analysis it is possible to approximate a definitive outcome. The Monte Carlo analysis or method rely on repeated random sampling to come up with a definite result.It uses a combination of probability andf statistics. Why is it important, is it because it allows us to simulate a definite outcome? No. Its important becasue it reminds us that computational algorithm is not reality but an approximation of reality.

Let me qoute Pablo Picasso to make sense of it all who said "If I know what I am going to do, what's the good in doing it?"

If most people are aware of the Monte Carlo analysis why is it then that 20% of the world's population holds 80% of the total wealth? Also why is that 20% of effort produces 80% of results while 80% of effort only produces 20% of results? Well we have Vilfredo Pareto to blame for that. However the Pareto principle can not be applied to everything but anybody who wants to plan their time optimally should know that roughly 20% of the time spent in a task leads to 80% of the results. That reminds me Louis E. Boone once said "I am definitely going to take a course in time management... just as soon as I can work it into my schedule."

Having said that the Pareto principle is not always right as observed by Chris Anderson former editor in chief of Wired. In 2004 he claimed that nearly everything that is offered for sale on the internet is actually sold as written on his blog. This is best illustrated using a demand curved called long tail model. In the model the best performaners account for 20% of the market, then the curve gently levels out which occuppies a much wider comprising of the less popular products but in todal bigger in number. The things is it still looks like the pareto principle applieshowever the long tail achieves a much higher total turnover than the best sellers.

If Chris Anderson is right, then don't worry if you don't end up with a blockbuster product, you might still end up with higher revenue. And if he is wrong, then you realize that Microsoft, Google and other companies like them are indeed one in a billion.

Monday, April 11, 2011

Single Platform for Network Management and Optimization.. What is it all about?

In the past month there have been heavy chatter on single platforms for managing various networks. This is focused mainly within the wireless industry, however there are a few industries that uses the term such as in the restaurant business or about computer OS as written by Hugo Ortega a few days ago or blogging capabilities. Six years ago there was discussions in using single platform databases for SOX compliance at the same time that the NoSQL movement started appearing to cope with the demand and specifically driven by the growth of social networks.

I would like to focus on networks, in particular wireless networks since it is an area where there are multiple vendors, platforms and solutions used by a network operator. In a typical wireless network operators, let's say a network with GSM (2G), UMTS (3G), LTE (4G) in its operation will have more than 3 RAN and CN vendors, most likely more than 10 transmission network vendors... and more vendors relating to operation support systems, business support systems and many other parts of an average operator organization. In most cases today each vendor will have its own way of managing and reporting their part in the network. The operator then is left with putting it all together from management, operations, reporting to optimization. The challenges here if not the problems are:
• Normalizing the real network performance across multiple vendor platforms.
• Matching the detailed technical reports to management reports
• Finger pointing amongst the vendors on where the discrepancies are
• Correctness of information, hence network performance
• Heavy resource requirement to get the job done


In the last two years the preferred solution seems to be having one platform to manage network performance and feed that to network optimization with the ultimate goal of Self Organizing Network (SON). The single platform then is supposed to gather data from various sources such as network recordings, NMS logs, OSS performance, drive test, probe logs, configuration dumps, and others. The platform then should correlate all this data, generate reports and if tied to SON, optimize the network. There are challenges of course in having a single platform:
• Accessing all data sources with various formats either directly or through repositories. Most vendors seem to like changing their formats for every updates they do.
• Large volume of data to be parsed, processed, correlated and reported
• Constantly evolving requirements by an operator based on new services, new technologies and subscriber development


There are a few solutions out there for a single platform such as ActixOne from Actix, Wireless Explorer's Unified Platform, and a few others. The question is, does it work? Are there other single platform systems available today?

Do you think it’s possible to implement a true single platform? What will it take?

Monday, April 4, 2011

Working in Commercial Flights... Does it Work?

Last week I flew to Kuala Lumpur from Copenhagen... that's a 12 hour flight and I had economy seats on a Singapore Airlines Boeing 777 plane. Since the flight departs morning I planned to work while on the flight... to my dismay there is not enough space to open my laptop and be comfortable enough to work... well at least type using the keyboard, and end up not being productive. When I got to Kuala Lumpur I felt like I just lost a day of work, and was very tired and sleepy on my way to an afternoon meeting.

On my return flight it’s the same story. It seems that on this economy flights there is not enough space to work, not even to watch an in-flight movie comfortably. In European flights there is enough space, at least, between seats that it’s possible to work while on transcontinental flight the space is tighter, plus the guy in front reclines his seat to be comfortable at your own expense. Now the question is can you really work while inside an airplane?

In business class or premium economy (old business class seats) for long haul flights, YES! Since there is enough space... In European flights, while in economy yes if you have a tablet pc or iPad or a small laptop. For long haul flights in economy forget about it, it does not work! This of course begs the question that "is it worth flying business in a long haul flight?" To me as an executive it is a decision based on cost effectiveness.

Here's how it works. If you fly for 12 hours, then potentially you can have 8 productive working hours. If your salary for 2 days is more than the cost of a business seat for your flight then I suggest you fly business since your productivity is more important than 2 lost days while on a plane and while recovering from such a long cramped flight. For short haul flights like European flights where you have enough space to work, better fly economy and by the way get a small pc if not an iPad.