Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
SUBMITTED ON
By
XXX
Sec-
Roll No. YYYYYYY
Professor:
CAN SPAM
On: yy-XXXXXXXX-zzzz
By
XXXX
VP, Communication Business Unit,
Ittiam Systems
E-Mail: yyyy@ittiam.com
INTRODUCTION
PROJECT DESCRIPTION
The biggest challenge in blocking mobile-phone spams is that this cannot be a content-
based technology, since there’s no call data till the time the user actually answers the call.
There are 2 ways to deal with it:
1. Reactive System - Setting up spam filters at a user level, where the user can
choose to block specific phone numbers.
2. Proactive System - Doing a pre-analysis at the network level to statistically and
probabilistically identify a spam call or message and drop it before being send to
the user.
The reactive method is like a fingerprinting based system and here the user is always
trying to catch up with the spammers. We have been working on developing a proactive
system that uses Bayesian probabilities to statistically identify a call or an sms as spam
and drop it before being send to the user. The basic trouble with using a proactive system
is that of the false positives, if the system wrongly identifies a call or a sms as spam, then
that message will get dropped, therefore we bias the probabilities slightly to avoid false
positives. The system works by using k-means clustering technique to slot calls or
messages into various types of baskets, where each basket is linked to a degree of
confidence and then depending on the user’s security level settings, the call is either
dropped or forwarded to the user. The system has been devised as a genetic algorithm,
which is self-learning and continuously imbibes all the data collected on the network to
fine-tune the cluster baskets. We have been testing our algorithms on a virtual test
network and the results have been shown in the annexure.
Keeping in mind the exponential growth in the spams on the cellular network, it is
imperative that the operators will have to start looking for solutions. Not only do the
spams annoy the end user, leading to decreased satisfaction levels with the service, it also
jams the network and decreases it’s useful throughput value. In view of this we propose
to build a system for Hutch based on our CAN SPAM algorithm, which can proactively
detect and drop spam data messages and voice calls.
It is proposed that the entire project will be divided into the following phases.
Phase I: Based on the requirement specifications, we make a detailed design and once
approved, build and test the system.
Phase II: The system is deployed on a small portion of the Hutch network, and any
corrective changes if required are made. Once the system is stable, the results are studied
for a period of 2 weeks.
Phase III: Hutch announces the feature to its customers and the system is officially
deployed on the Hutch network.
Phase IV: We provide full support for 2 months from the deployment date and handle
any issues concerning the system. We will also survey consumers and understand the
consumer acceptance of this novel feature. We plan to do a do a dip-stick survey among
2000 Hutch users.
Detailed Desig n
Deployment
Project Management
Support
Note: Execution of activities may overlap in time, as shown above
DAY MILESTONE
1 Project Initiation. Signing of the contract
5 Functional Specifications have been frozen
MUTUAL RESPONSIBILITIES
We would require the following support from Hutch to ensure successful project
completion: -
• Clear and unambiguous definition of the project scope and deliverables.
• Mentoring session by the company’s technical division for increasing our
understanding of the Hutch network.
• Access to information wherever required and relevant.
• Review by an identified Hutch personnel after every stage of completion.
• Access to Hutch’s intranet and necessary project servers for the Ittiam onsite
coordinator and Ittiam offshore development team.
ACCEPTANCE PARAMETERS
• The throughput of the network should not decrease because of the CAN SPAM
system.
Both the parties would need to mutually agree upon the following performance levels
of the system: -
• Percentage of spams detected
• Percentage of false alarms raised by the system.
COMMERCIAL TERMS
The cost is on Time and Material basis. The total cost involved in the project is as worked
in the appendix.
The total estimated project cost is Rs. 25,51,655 (including contingency costs and
royalty).
Incidentals would be based on actuals.
Payment Terms
Ittiam Systems follows a policy of minimum credit. A contract-signing amount of Rs
2,00,000 is to be paid after which the project will start. This will be adjusted in the first
phase installment. Payment terms/amounts initially agreed upon would be non negotiable
once the project commences. All the payments for the completed phases need to be paid
before starting the next phase. Payment would be done after a common review of the
deliverables has been done and a mutual agreement arrived at. The contingency cost and
the royalty cost is to be paid within 15 days of finishing the Phase III.
CONCLUSION
Spam data messages and voice calls are exponentially growing, making it not only an
annoyance but also possibly a threat to the continued usefulness of the mobile phone
itself. Ittiam Systems has developed and patented the algorithm for spam detection and is
at least 6 months ahead of the competition in research. The CAN SPAM project offers
Hutch a USP to offer to its customers. Also it will enjoy the first mover advantage in the
market. As a byproduct of spams being dropped from the hutch network, the useful
throughput of the network will increase considerably leading to faster connection
responses to its users.
The estimated completion time of the project is 7 weeks and we also offer 8 weeks of
post deployment support. The estimated project cost is Rs. 25,51,655.
Annexure
Founded on January 1, 2001, Ittiam Systems is a DSP systems company. Our mission is
“to be the World's best DSP Software and Systems Company”. We have competencies in
multiple end equipment domains, covering Wireless, Wireline, Speech/Audio, and
Imaging/Video. We undertake product development for customer specifications, covering
all or part of the steps in the overall product development cycle. Ittiam Systems is led by
a team of professionals with over 130 years of experience in the management,
development and marketing of high technology products. We are headquartered in
Bangalore, India, with offices in the US, UK and Taiwan and representation through
authorized Associates in the US, Japan, Taiwan, Korea and Singapore.
YYYY who has over 14 years of industry experience will lead the project. He is widely
renowned as an expert in Data mining and genetic algorithms. He is a member of the
Asia leadership team and a member of the Karnataka State IT Task Force.
ZZZZZZZ will be the technical lead for the project. He has over 6 years of industry
experience and has been the chief architect of the CAN SPAM algorithm.
PRICING DETAILS
Table 2 Rates
Role Rate/Hour
Project Manager xxxx
Development Engineer yyyyy
Test Engineer zzzzz
Deployment Engineer wwwwww
Support Engineer uuuuuu
Table 2 Costs
No. Of hours Total Cost
Phase P.M. Dev Engg. Test Engg. Dep Engg. Supp Engg.
I
II
III
IV
Contingency z% of total
Royalty for CAN SPAM algorithm
The total project cost is Rs. yyyyy (including contingency costs and royalty).
In case any installment is not paid by the date on which it is payable, a notice may be
served to pay the installment within a period of two months together with a penalty which
may extend up to y%of the amount due.
TEST RESULTS
ASSUMPTIONS
• Hutch will not sublease the CAN SPAM network to any other cellular operator.
• Ittiam will not use the intrinsic knowledge gained about the Hutch network
outside of the project.
• Guaranteed results in terms of 100% spam detection rate and 0% false alarms are
out of scope.
• Acceptance tests will be performed only by Hutch personnel, Ittiam will only
resolve/fix any issues identified.