Internet of Things IoT

[yorkie820]

Hello,

How does everyone else feel about the IoT movement? I see it as the future of BAS and controls..... any thoughts?

[loopypid]

Based on previous threads, it appears your view would be in the minority: http://hvac-talk.com/vbb/showthread....s-quot-article

I agree with you that it's absolutely the way the world is moving. Call it what you want, "intelligent building," "Building Internet of Things," etc. - it's coming, and building owners are actually starting to pay for it these days whereas before, this was more of just a pipe dream.

[orion242]

Little more than marking wank IMO. Just try and find a definition of IoT.

We have been putting systems on the net for more than a decade, whats new with IoT?

Sure they have all these ideas that big data is going to change everything, yet find one detailed case study showing they did something new (that couldn't be done previously) and had real ROI. You have one camp running around calling for more security on these systems and the IoT guys trying to get everything opened up to 3rd parties via the internet.

All that aside, none of their promises can happen without crap loads of labor to put meaningful metadata on all the information so they can actually analyze it. How many systems are using haystacks or similar today? 0.1*10^-25% of all installed systems? We are using haystacks and looking at a few of these packages today, none I find all that impressive for the labor required and the upfront & reoccurring costs. Haystacks itself seems half baked. Start using it and you quickly notice they are missing tons of what should be pretty common tags. N4 has its own tagging system which almost nobody but Tridium's own analytics package uses. Multiple tagging systems should help the adoption rate...

Will it be a fad like 3d TV or something with real legs? Who knows, but it certainly looks to be a long bumpy road littered with bull$hit & promises at the moment.

[BetterDuck]

I don't even understand why anyone needs to bring up Haystack. It is a manufacturer-american-centric creation. This data gathering still needs many things to make it "go".

The issue not many address with IoT is how all these devices interact with each other and people.
Some companies have some ideas, and less have actual implementations.
But, those implementations are working. And how? Well, you have a group of devices in a cloud and they know how to talk to each other. They know how to tell other dependent devices if their data is good or they should be evaluated slightly different. They even have a complete functional description so servers can visualize their control scheme. The exact same control scheme can be put into multiple manufacturers devices.

Currently, with systems people manually make associations (e.g., client mapping) and then all the programming and network tools are per each manufacturer. Of course the biggest problem is most manufacturers don't even have a decent lineup of IP controllers either.

[orion242]

And in walks yet another competing solution...here to save us where all else have failed.

Any detailed white papers with real data on the roi or just plastering more marketing wank into the mix?

[BetterDuck]

And in walks yet another competing solution...here to save us where all else have failed.
Any detailed white papers with real data on the roi or just plastering more marketing wank into the mix?

Some days I just cannot avoid responding to what I think is garbage.

IEC is a non-profit, non-governmental international standards organization that prepares and publishes International Standards for all electrical, electronic and related technologies. In other words, the opposite of proprietary structure-frames or stacking hay initiatives some people can't stop genuflecting about. The IEC has a standard 61499. And in that IEC standard is much of the "glue" that IoT devices actually need and can use.

Yes, there is detailed white papers from non-profits (not like stacking hay or proprietary structure-frames), independent university research (not like stacking hay...) and other non-governmental groups.

Any detailed white papers with real data on stacking hay, or structure-frames, or "just plastering of more marketing wank into the mix?" I recall the CEO of some proprietary frameworking company at their summit semi-seriously asking for customers to document their platform. Did they pull that video off the net?

[Frozenlock]

I'll pitch in and say that the amount of coverage and publicity Haystack receives is absolutely not proportional to its quality.
(I've written a review a year and a half ago.)
The structure looks like someone just took an old-ish proprietary stack and added a little paint to make it look new and shiny.

I honestly don't see IoT coming shortly.

The issue not many address with IoT is how all these devices interact with each other and people.

Exactly.
There's simply too many "things" around to build a universal ecosystem.
How would a shoe announce itself on the network? A cat litter? A TV?

Finding a way to intelligently describe what each device does and giving (restricted?) access to its sensors and outputs is really, REALLY hard.

If you are simply talking about a subset of a domain, say for example "controls", then you are not talking about IoT, but rather another standard.
We already have those. They are clunky and bloated, but they work.

It's not clear to me that yet-another-controls-standard would bring to the table... but I'm open to new things, if they can prove worthwhile.

[orion242]

And in that IEC standard is much of the "glue" that IoT devices actually need and can use.

Its a programing standard, bit of a stretch to apply it to data modeling. How many 3rd party analytics packages are out there that can use it in any meaningful way?

there is detailed white papers from non-profits (not like stacking hay or proprietary structure-frames), independent university research (not like stacking hay...) and other non-governmental group

Yep, I looked into them a while back when some other forum member kept mentioning how wonderful it was. Dig past the wank, and you quickly find plenty of papers describing how 61499 doesn't live up to its expectations. Since manufactures can implement custom blocks, guess what...still not using vendor's A software and programming and dumping it right into vendor's B hardware. The more things change, the more they stay the same.

So using your 61499 compliant tool how many 3rd party hardware bits actually work with it?

Seeing how you cling onto lon so strongly, somewhat surprised your not behind Echelon's IzoT platform.

[orion242]

I'll pitch in and say that the amount of coverage and publicity Haystack receives is absolutely not proportional to its quality

This is the hallmark signature of just about all things IoT. Tons of polished marketing telling you how great they are, while just under the covers there is little if any substance.

The only thing positive I can say about haystacks is at least there are a few packages that can actually use it. I don't see it really gaining a heck of alot of traction just yet. Could easily be here today, gone tomorrow.

[BetterDuck]

I'll pitch in and say that the amount of coverage and publicity Haystack receives is absolutely not proportional to its quality.

Congratulations... that's the understatement of the year. Your review was excellent. My opinion is that it would be nice to get an Identification Key (Siemens derived?) into a property of a BACnet object and as long as you have the solution you now have a tag going down to every point in a system. And, that is compact.

The things I like about the control standard is the portability of the code and inherent connectivity. This eliminates the error of devices interacting with each other and providing good data or denying data which is not good. Another mentioned three basic things regarding 61499:

• portability: Software tools can accept and correctly interpret software components and system configurations produced by other software tools.
• interoperability: Embedded devices can operate together to perform the functions needed for distributed applications.
• configurability: Any device and its software components can be configured by software tools from multiple vendors.

It is my opinion that IoT devices would need these elements in some form to assemble relevant data between them.

[orion242]

get an Identification Key (Siemens derived?) into a property of a BACnet object and as long as you have the solution you now have a tag going down to every point in a system.

From what I understand Structured View objects exists in the bacnet standard now with the idea they can be used to define the relationships between points, etc. Can't say I have really dug into them and since no one supports them...its all pretty pointless at the moment.

[BAS-atechs]

How does everyone else feel about the IoT movement? I see it as the future of BAS and controls..... any thoughts?

Thanks for posting this, yorkie820, and also to the members replies. I've been reading lots of articles recently about IoT and Haystack, which mostly sound optimistic. So it's good to hear opinions on forums like HVAC-Talk to give an even broader perspective of this subject. Please keep them coming!

[BetterDuck]

...Since manufactures can implement custom blocks, guess what...still not using vendor's A software and programming and dumping it right into vendor's B hardware. ... Seeing how you cling onto lon so strongly, somewhat surprised your not behind Echelon's IzoT platform.

XML is accessible and readable. 61499 blocks must be stored in a XML format. Secondly, if you don't like it write your own and execute that code on the product. There are examples of this out there including SCADA which interprets the graphics in the block.

Of course everything could be better but there are a bunch of people running around that don't even know what to ask for, specify and demand. They would rather complain about nonsense and take what is fed to them. This is why more specifications should be performance based and then a performance score be assessed for compatibility. Too much to handle for most specifiers when they can more easily write a few manufacturer names in a list.

Surprised on IzoT? This is an example of not paying any attention all these years. The structure of LON and FT-10 was compact and concise. Coupled the Manchester encoded and collision avoidance network but it way above in performance over MSTP, especially the disaster BACnet MSTP was in the early 90's. That's it. Now we have IP and it carries the heavier nature of BACnet, albeit without anything reasonable for security. Both of which are ISO standards. Is your favorite structure-frame an ISO standard? .... Not a chance. The spawned control logic an IEC standard or any recognized and supported open standard? Again, no. And, you support those things over international standards so you are part of the problem.

[yorkie820]

Thank you so much for the support and endorsement. We need more forward thinkers that can embrace the future. We need less doubters and less negativity.

[BetterDuck]

Its a programing standard, bit of a stretch to apply it to data modeling. How many 3rd party analytics packages are out there that can use it in any meaningful way?

No stretch whatsoever on data modeling, but I'm not going to give up what I know and see for that.

What I will point out is that most analytics is also garbage and riding on top of a basis of a weak offering of building automation processing power from fieldbus devices running extremely limited data rates.

For instance, let's say you have an IP ASC controller with 4 million trend entries and more processing power than a standard middleware heavily licensed box with a proprietary communication structure-frame. Now, you put 200 of those in a building. They are doing their own analysis of their functionality because there is all sorts of processing power available. No more Viagara-network which gets you up momentarily and eventually lets you down when things get serious. Start serving up compiled data. Why would you want to collect 15 minute energy intervals when each controller tells you various totals like last 7 days, month and YTD? Why grind that data in a secondary database somewhere when you can receive it as a single data point from the controller?

Changes this analysis game entirely. It becomes a results game at that level and changes how much you need to analyze in a central database. You can also distribute the queries in the network rather than crunch them in a big data base. It is not eliminating all global queries, but anyone can see significant gains with the distributed processing power. This company called Google knows about this and people can research this.

With all that distributed processing power each box can contain a copy of the entire site monitoring graphic or other types of graphics which changes entirely how an individual looks at deploying buildings. Currently you need to go to the box or server for everything. Why do you need expansive logins and heavy capacity at a central server to run a room and set some offsets? Wouldn't it be easier to walk in a space and quickly connect to the controller in the space and get that general information?

That's some IoT and most manufacturers are now almost stumbling over themselves on how to get there as quickly as possible.

[orion242]

Well we can listen to a bunch of techno babble dodging real world questions and assume it must be great. Better memorize it though. At some point you'll have to pull the same smoke screen when your customer is asking why things are not as rosy as claimed.

If you actually have some understanding of the technology, go out and hunt these vendors down. Have them show it working with your offerings. Figure out, how much labor you’re going to have to implement their products, how much its going to cost on a typical site, what are the training costs, etc. Can I use it with other vendors platforms. Ask them for a few customer references and talk to them. Do the customers feel like they got real value, what do they like about it, what don’t they like? Do you have enough customers that will actually step up pay for these services to make it worthwhile to put your sales team on it?

By the time you dive in that deep, you should have a pretty good idea of the reality of IoT as it stands today.

[orion242]

Refreshing to see you can ditch the babble for a min.

What I will point out is that most analytics is also garbage and riding on top of a basis of a weak offering of building automation processing power from fieldbus devices running extremely limited data rates. .

The majority of devices out there now have less than 15-20 points on them. Some portion of that will be digital points, ie 1 bit values. The analog points at best are 12-16bit values. None of this I need to updates more than maybe once a min, on the very high end, to run meaningful analytics. Space temp doesn’t move fast. I don’t need 100ms updates, it’s POINTLESS. Less than once every 5min is likely useless, not process control. Give me analytics with 5-15 min reaction times, more than sufficient and possible even on the slowest POS fieldbus. Even faster on the points that matter is little problem for the vast majority of the fieldbuses installed today.

Why the do I need high bandwidth exactly?

What do you expect your market is when you calling for a complete rip out and replacement of a BMS just for your setup? There is no reason current devices can't be overlaid with some serious smarts now.

let's say you have an IP ASC controller with 4 million trend entries.

Why wouldn’t I be doing bulk storage somewhere central? One fast machine, or a hierarchy bottom up process. How are you doing long term archival and backup when it scales up? How is this going to be more cost effective than edge devices holding a few weeks and pushing up to a datacenter where now it’s a single machine, or farm of machines trickling up data? Single db query, or pull from crap loads of devices...

let's each controller tells you various totals like last 7 days, month and YTD? Why grind that data in a secondary database somewhere when you can receive it as a single data point from the controller? .

Again, processing power in PC is a commodity item and piss cheap. Even in this model, I have to pull all this crap from hundreds or thousands of sources to get floor, building, campus wide totals. What about worldwide accounts? How are you going to scale this? IT knows managing servers and it’s easy to scale, replicate, virtualize etc.

let's Changes this analysis game entirely. It becomes a results game at that level and changes how much you need to analyze in a central database.

Explain. As I see it, my VAV is a drop in the pond. I want to work with all the zones and optimize the central plant. I want to look at data from 50k feet and see patterns or make changes the affect central plant. The edge devices need to dance to my tune, but the music is written looking from above for the greater good. Your model, my garage door light bulb should be telling the power plant what’s best without seeing what’s going on in grid as a whole?

let's Why do you need expansive logins and heavy capacity at a central server to run a room and set some offsets? Wouldn't it be easier to walk in a space and quickly connect to the controller in the space and get that general information?

And what’s the user experience if they have pop in wads of these devices? So what happens when the energy team needs to look at multiple buildings, zones, etc. They logon once and it’s good throughout the org? How is user management handled, and most importantly how is user authentication passed between devices? Sounds messy at best.

Heavy capacity? How many concurrent users are you dealing with? Facebook, google, etc. This has been solved with commodity hardware that IT is familiar with. How many times do you login with google across their properties that spans server farms? Even the largest sites I have seen deal laughable numbers compared to what current solutions can deal with.

Want to get man on the mars fast?

Tell IT you need high bandwidth connections to every device org wide, it needs to be open to some 3rd party, it’s their baby to manage every device and their budget gets to pay for the infrastructure.

[orion242]

BTW...

Control Talk last webcast...Ken Sinclair mentions that these edge devices are apparently available now for well < $100.

So who is offering any UL listed, commercial device for the prices anywhere in the universe he talks about??

Someone needs to check his temperature, I think he is delirious and in need of medical attention. This happens with two distributors of controls on the line who are well aware of actual costs just eating it up like its reality.

[orion242]

https://youtu.be/gjJInSMkSMs?t=13m38s

$50 even $30 for a controller with cloud and all this wiz bang edge power available now!!!

This couldn't be further removed from reality.

[kontrolphreak]

Just like the majority of the posts in this thread, IoT and analytics are a joke as currently deployed / envisioned.

kontrol out