How IoT will Dis­rupt En­ter­prise Com­put­ing

DQ Channels - - Channel Pulse - SHRIKANTH G/(shrikan­thg@cy­ber­me­dia.co.in)

Ganesh Moor­thy–Ap­pren­tice Leader, Mu Sigma, talks about IoT evo­lu­tion and its im­pact on en­ter­prise IT or­ga­ni­za­tions

IoT is the new rage. There is a huge op­por­tu­nity wait­ing to be tapped. A Gart­ner re­port re­cently said that 6.4 bn ‘con­nected things’ will be in use world­wide in 2016, up 30% from 2015, and will reach 20.8 bn by 2020. In 2016, 5.5 mn new things will get con­nected ev­ery day. Gart­ner es­ti­mates that IoT will sup­port to­tal ser­vices spend­ing of $235 bn in 2016, up 22% from 2015. Ser­vices are dom­i­nated by the pro­fes­sional cat­e­gory (in which busi­nesses con­tract with ex­ter­nal providers in or­der to de­sign, in­stall, and op­er­ate IoT sys­tems).

With this back­drop, ven­dors across the IT spec­trum are ink­ing ag­gres­sive IoT strate­gies and it rep­re­sents ma­jor rev­enue op­por­tu­nity. Let’s look at Mu Sigma which is mak­ing defnite in­roads in this seg­ment. Mu Sigma de­fines it­self as a cat­e­gory-defin­ing de­ci­sion sciences and big data an­a­lyt­ics com­pany, help­ing en­ter­prises ‘sys­tem­atize’ bet­ter data-driven de­ci­sion mak­ing. The com­pany be­lieves that its in­ter­dis­ci­pli­nary ap­proach and in­te­grated ecosys­tem of plat­form, pro­cesses and peo­ple are re­defin­ing how com­pa­nies ap­proach prob­lem solv­ing in ar­eas of mar­ket­ing, risk and sup­ply chain gives it a dis­tinct edge in the mar­ket. With more than 3,500 de­ci­sion sci­en­tists work­ing across 10 in­dus­tries Mu Sigma says that it has been con­sis­tently val­i­dated as the pre­ferred de­ci­sion sciences and an­a­lyt­ics part­ner for large en­ter­prises. In an in­ter­view with Dataquest, Ganesh Moor­thy–Ap­pren­tice Leader, Mu Sigma, talks about IoT evo­lu­tion and its im­pact on en­ter­prise IT or­ga­ni­za­tions. Ex­cerpts:

Can you talk about the multi-pronged im­pact of IoT and how it will dis­rupt en­ter­prise com­put­ing?

As de­vices get smarter and han­dle more com­put­ing on the edge, the vol­ume of data be­ing routed through the en­ter­prise net­work will drop. But as IoT de­vices be­come ubiq­ui­tous, over­all data vol­umes will in­crease tremen­dously. En­ter­prises will have deal with this del­uge of in­for­ma­tion at scale and in­creas­ingly in real time. Ma­chine data will drive even greater de­mand for big data stor­age, and new, more adapt­able scale out mod­els for server farms. Now add in the wide va­ri­ety of com­mu­ni­ca­tion pro­to­cols, and com­plex­i­ties in ar­eas of net­work man­age­ment and in­for­ma­tion man­age­ment are com­pounded.

Ow­ing to var­ied lo­ca­tions of de­vices (within or out­side of the fire­wall) net­work se­cu­rity will be crit­i­cal and will re­quire new poli­cies be put in place. Analysing dis­parate data will be­come the main fo­cus to­wards in­sights for pre­dic­tive and/or pre­ven­tive pur­pose. En­ter­prises will need to in­vest on real time com­plex event man­age­ment sys­tems to keep up with in­for­ma­tion stream.

Do you think IoT has passed the hype curve and on its way to adop­tion?

IoT has cer­tainly passed the hype and surely on its way to en­ter­prise adop­tion. We are see­ing a lot more use cases for IoT rang­ing from ac­cess­ing in­for­ma­tion from fit­ness de­vices to smart cities, where ev­ery de­vice is in­ter­con­nected for most op­ti­mized liv­ing con­di­tions. Telem­at­ics is prob­a­bly one of the big­gest use case right now.

How will an­a­lyt­ics aid in bet­ter IoT?

IoT and an­a­lyt­ics go hand in hand. Take fit­bit for ex­am­ple. While the de­vice it­self has gy­rom­e­ter, ac­celerom­e­ter and pulse mon­i­tor to mea­sure your ac­tiv­i­ties and heart­beat, you need a cor­re­spond­ing ap­pli­ca­tion to an­a­lyse the pat­tern. A more com­pli­cated ap­pli­ca­tion is fa­cial recog­ni­tion through videos. The video cam­era cap­tures live feed, which are de­coded and run through a num­ber of modelling tech­niques to de­tect fa­cial fea­tures for iden­ti­fi­ca­tion. We have been ex­per­i­ment­ing with de­vel­op­ing an­a­lyt­ics on small com­put­ers such as Rasp­berry PI and Ar­duino to­wards edge an­a­lyt­ics where most anal­y­sis takes place on the de­vice it­self. Re­sults are then sent to the server for his­tor­i­cal pat­tern de­tec­tion. We have suc­cess­fully used sim­i­lar ap­proaches to de­ter­mine on-shelf avail­abil­ity of prod­ucts in real-time, send­ing out no­ti­fi­ca­tions to ap­pro­pri­ate stake­hold­ers for re­stock. This ap­proach re­sulted in sig­nif­i­cant sav­ing in com­bi­na­tion with op­er­a­tional ef­fi­ciency and re­duced hu­man ef­fort.

What role is Mu Sigma play­ing in IoT?

Mu Sigma con­tin­ues to in­no­vate in the devel­op­ment and ap­pli­ca­tion of in­tel­li­gent de­vices that in­cor­po­rate edge com­put­ing for va­ri­ety of busi­ness pur­poses. We’re build­ing en­ter­prise sig­nal pro­cess­ing sys­tems to process stream­ing, real-time data and vi­su­al­ize in­sights in a way that makes in­tu­itive sense to busi­ness de­ci­sion mak­ers. We’re work­ing on stream­ing video data and an­a­lyt­ics ca­pa­bil­i­ties for large re­tail­ers and hospitality com­pa­nies, RFID sen­sor data and ma­chine level data from wear­able de­vices, us­ing a cus­tom big data plat­form built on JADE, Spark, and Storm. We’re also help­ing clients col­late ma­chine data from legacy in­dus­trial ap­pli­ances, and us­ing this in­for­ma­tion for pre­ven­tive main­te­nance pur­poses.

If you say that you are build­ing en­ter­prise sig­nal pro­cess­ing sys­tems—can you throw more light on the specifics?

Our en­ter­prise sig­nal pro­cess­ing sys­tem is a state of art frame­work to help build real-time com­plex event pro­cess­ing based ap­pli­ca­tions. This can be used to de­tect sig­nals from data com­ing from IoT based de­vices or any re­al­time feeds and per­form the nec­es­sary ac­tions as re­quired. This plat­form al­lows for users to build cus­tom event pro­cessers us­ing flow based par­a­digm and comes with highly con­fig­urable in­ter­face for front end vi­su­al­iza­tion. It sup­ports both par­al­lel and dis­trib­uted com­put­ing and scales on de­mand.

What ac­cord­ing to you are some of the in­hibitors on the road to IoT adop­tion?

While some seg­ments like man­u­fac­tur­ing have started adopt­ing IoT sig­nif­i­cantly, oth­ers are fac­ing chal­lenges un­der­stand­ing the busi­ness im­pact of IoT it­self. For those who do make a valid jus­ti­fi­ca­tion for IoT, lack of ex­pe­ri­enced skillset is the next big hur­dle as it in­volves deal­ing with hard­ware, em­bed­ded soft­ware and ad­vanced ma­chine learn­ing/deep learn­ing tech­niques some­times. Fi­nally, data se­cu­rity and pri­vacy con­cerns need to be ad­dressed as ma­chine-to-ma­chine in­tel­li­gent com­mu­ni­ca­tion be­comes ubiq­ui­tous in the years to come.

GANESH MOOR­THY AP­PREN­TICE LEADER, MU SIGMA

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.