How Technology Failed in Iraq

[Bartok5]

Well, I certainly can't disagree with anything you've said in your latest post.  Indeed, I fully agree!

Hmmm.... perhaps our views are not so disparate as I had originally thought 

Cheers,

 

[pbi]

Perhaps the model to persue is the US Special Forces in Afghanistan: Operators directly linked to the shooters without the intervening Headquarters apparatus. There, airpower could be vectored in to a target within minutes. Previous wars, like the Persian Gulf War and Kossovo had delays of several hours between the target being sensed, and an airplane finally arriving while the information worked its way up and down the chain.

a majoor: I see what you are driving at, but that isn't exactly how it happens here now. While CJSOTF (Combined Joint Special Operations Task Force-the main SF component of CJTF76) may have indirect fire assets cut to support it (esp to provide fire sp for the Forward Operating Bases) this doesn't really extend to air assets-they pretty well follow a centralized process (which works). As well, almost all of the activities of CJSOTF are briefed twice daily here at the CJTF76 HQ, on the "Big Screen", and are subject to the direction of the Commander CJTF76. Because of the concerns of collateral damage, and the danger of accidental strikes into Pakistan, there is higher control over what gets hit with air strikes.
I do know, however, that squad leaders and team leaders in conventional Inf units are taught to call for fire.

In my opinion, the USMC is probably further along (doctrinally, anyway) with "Sensor-to-Shooter" links than the Army is, if only because the USMC tends to function at lower levels and somewhat more decentralized. Cheers.

 

[a_majoor]

I have had the pleasure of working with the Marines, and yes, they do seem to be more switched on WRT "sensor to shooter". Marine Lance-Corporals are trained and exercised in the art of calling air and artillery strikes, for example. The book and movie "Blackhawk Down" shows the problem with centralization: orbiting aircraft could give the commander the "big picture" view of the battle, but the number of steps between the airplane and the ground convoy commander meant the "traffic directions" arrived to the convoy commander AFTER he had driven past the intersection he needed to turn at.

This may be more in tune with the "Army Structure" thread, but here is one idea: The "sensors" feed both to the HQ and the "shooters". The "shooters" job is to use the real time data to deal with the current engagement, patrol or whatever. The HQ's job is to interpret the data and do a "set up" of assets for the next projected evolution, so when the "shooters" need support, the aircraft are already orbiting overhead, reinforcements are arriving for the exploitation or the FLG convoy is rumbling up the road for the DP. The long loop for support is cut down considerably for the "shooters". HQ's should be able to listen into the "shooter's" nets, but be constrained by training and protocol from getting too active on those nets. Shooters should be getting the commander's intent, and updates on where the support is coming from ("change to channel 8 at 1300 for the AC-130 which will arrive on station 1300-1500...").

This isn't too different from what we do now, actually, but the two keys differences are direct feeds to the shooters, and forcing the HQ to take a hands off approach to the current situation and be more proactive in the "branches and sequels" aspects of planning.

 

[RDA]

This may be more in tune with the "Army Structure" thread, but here is one idea: The "sensors" feed both to the HQ and the "shooters".

This idea is reminiscent of a problem that was faced during the mid 1980s in the field of robotics.   The following is cut-and-paste straight out of a paper I wrote a few years back on the topic of aerial robots (which are essentially UAVs with more autonomous capabilities).  
Perhaps some insight into the problems created by technology could be found within the technology itself:

Three popular architectures have been widely documented in the study of goal-oriented robotics. The first is a two-layer design known as the â Å“Sense-Model-Plan-Actâ ?, or SMPA approach. In this system, the upper-layer uses the robot's sensors to create a model of the environment, and then uses this model to plan a course of action. The upper-layer then sends commands to the lower-layer, which executes these commands blindly. While this approach can be adequate under static conditions, its effectiveness rapidly deteriorates as the environment becomes more dynamic. The weakness of the design lies in the computational complexity of the SMPA algorithm, which introduces unwanted system latency. The loss of performance occurs as this latency becomes too large for the robot to react quickly enough to its changing environment.

The second method used in goal-oriented robotics is a â Å“behaviour-basedâ ? approach, whereby a complex task is made more tractable by decomposing it into many relatively simple primitive behaviours.   Each of these independent behaviours reacts to the raw sensor data by issuing commands that would best achieve its own narrowly defined objective.   The commands generated by all the primitive behaviours are then fused together in real-time using one of many possible arbitration strategies, and the resulting commands are sent to the robot's actuators.   While this type of system, also know as a â Å“reactive controlâ ? system, works well in unknown and/or transient environments, it cannot generate plans from goals or make use of a priory or learned knowledge to make decisions.

The third and most sophisticated architecture is a combination of the first two approaches, developed in an attempt to overcome the reaction-time issues of the SMPA architecture by augmenting it with the rapid reaction capability of behaviour-based systems.   This design, known as a â Å“hybridâ ?, adds what can be described as a â Å“reflex unitâ ? between the sensors and the execution layer of the standard SMPA controller.   While this architecture does nothing to reduce the latency in the primary control loop, it adds a secondary and much faster control loop whose purpose is to provide reactive control in emergency situations where the time required to first model (map) the environment and then plan (compute) a response is not available.   In such situations, the robot must simply react to the raw sensor data if it is expected to survive.

 

[Evocatus]

To: Mark C,

The original article:  How Technology Failed in Iraq

Is actually posted as well to :

http://www.iraq.net/displayarticle5830.html

"iraq.net" surprisingly.


A very interesting article for those in the Signals and Intelligence
profession, specifically. that relates very closely to your observations
about Afghanistan: Below Div level, all the
new shiney bells-n-whistles like ISTAR, GPS etc....is not currently, a
reliable  asset.  Along with the old and ever current philosophy; we had
better have our basic soldiering, nav and AFV recognition skills up to
scratch, regardless of how technologically advanced we and/or our
intelligence gathering assets become:

=================================

 

[54/102 CEF]

Sensor to Shoot-ee is my view - http://users.ev1.net/~medosier/Webpics/Bin-Laden-Priceless800x600.jpeg

I attached the picture in case it is taken off line

No matter how the eqpt evolves there will always be the few hundred yards between you and them.

As for the politiically driven procurement system - not much new stuff there. Nor is the lack of senior officer resignations.

Next stop - hell.

But isn't that what war is?