User in the loop

from Wikipedia, the free encyclopedia
The UIL concept integrates the user of a system in order to achieve performance goals. Occasionally, information or instructions are sent to the user by a controller.

User-in-the-Loop (UIL) is the principle that a target (e.g. improving the efficiency of a network) is achieved by influencing the users involved. The principle can be applied in many different disciplines. One assumption in UIL is that the human user is intelligent, but his behavior (output) is much less predictable than the input / output behavior of technical systems . Furthermore, users also have a limited amount of input variables that are carefully followed by the senses. That can u. a. visual stimulation, acoustic signals / speech / music or even haptic effects (imagine an accelerator pedal that offers a force as resistance when leaving the efficient operating range). Both elements, the intelligent decision of the user and his observed input values, can be used to achieve the great goal. The purpose of the input values ​​is to motivate the user to a certain behavior or to advise against it. The desired behavior can only be seen in relation to the goal and should not affect the entire action.

A related example from the past is a tiering of prices for energy prices depending on the use of electricity at peak time or off-peak time. If this information changes dynamically and not at fixed times and if this information is communicated to the user promptly (preferably before use), then it is already a controlled user. In this example the control is open (without feedback). UIL allows a real regulation (closed loop control), i. H. with the user IN the control loop.

If users are confronted with a dynamic price (e.g. for electricity ), they react as expected by adapting their usage behavior, which leads to an improvement in energy use by all users (reduction of peak load ).

In addition, UIL was proposed for cellular communications to avoid capacity bottlenecks.

In this application, radio resources are scarce . In the past, systems could be oversized (demand is less than supply). However, it has been shown that telecommunications networks are recording an enormous exponential increase in data rates, which, according to prevailing opinion, cannot be caught up by any technological development. The capacity crisis is thus inevitable. End users will notice this e.g. B. through broken connections or slow videos.

While everything is being done on the technical side to provide more capacity (e.g. with better resource allocation, MIMO, cognitive radio, machine learning), it looks as if the main cause of the problem is being ignored, namely the user himself. In wireless networks users can actually be influenced to change their usage behavior by introducing incentives (e.g. dynamic prices).

In addition, environmentally relevant goals can be achieved using UIL. In the area of ​​radio networks, less usage also leads to a lower ecological footprint (since every bit sent requires a proportionate amount of energy).

UIL for (radio) networks is also called the smart grid of communication. It aims to avoid unfavorable locations (if the sewer condition is poor) or against use during rush hour .

Overview

Regardless of the possibilities of giving incentives and sanctions, the system response of the user block (in the context of the communication system ) is either a spatial, temporal or no response. Spatial UIL means that the user changes his position towards a better one (a common practice in the Wi-Fi network ). Temporal UIL means that the user changes his request or waives it at the moment (in order to continue at another time, to give up completely, or to satisfy the need in some other way, e.g. via cable or Wi-Fi at home). The incentive (entry into the user block) is usually a fully dynamic price ( mobile phone tariff ). This changes the usage behavior during traffic jams (see Network congestion avoidance ). UIL aims to stabilize the traffic demand so that it is at a sustainable level below the capacity limit. In cellular networks ( mobile communications ), this measure helps to keep traffic below capacity at all times (medium and long-term). There are always short-term congestion situations in networks (see Best Effort ).

Spatial UIL scheme

The general principle of UIL can be seen in the picture above. In the UIL paradigm, the controller gives the user the necessary information, and it is expected that the user will voluntarily change his position from A to B. The current reception quality ( signal-to-noise ratio ) at point A and the associated spectral efficiency are known to the algorithm. In addition, the average values ​​(and if necessary more statistics) of the above metrics for all relevant locations (including B) of the network are known (from a database condensed from measurements from the past). After a search, the algorithm prepares the required information and suggests one or more better positions to the user. Before moving, the user knows his benefit or incentive of position B compared to A. This incentive can be of a financial nature (e.g. discount for a telephone call) and / or an improved data transmission rate for best effort data traffic. The server then provides information about where (in which direction and to which place) one should move. Before making a decision, the user should have all the information (discount, improvement of the data rate, how far is it to point B). Eventually, a certain percentage of users will follow the proposal and the rest will remain unchanged (this includes all users who cannot or do not want to move, or whose incentive is insufficient). In the system diagram, the user's system block outputs the new position B when the user moves. The probability depends on the distance and the incentive. The target value is the minimum required spectral efficiency that the user should achieve after moving (this target value at B must be greater than the current value at A). Typical distances are roughly on the order of ten meters.

Temporal UIL regulation

The increase in traffic demand in mobile communications is fueled by flat rate tariffs. The result is a heavy-tailed distribution of the monthly data volume per user. It leads to unlimited exponential growth . It has been observed that tariffs are changing because of unrestricted growth. Some Internet service providers and telecommunications network operators have started to offer flat rate tariffs with a cap, but this is only a temporary solution. To speak of a flat rate when it comes to a fixed volume is also not fair to the customer. A more sensible solution is usage-based billing. The customer's fear of this is only due to the bill shock. H. Tricks used by operators to debit a disproportionate amount of money in certain situations, but this can be countered by transparency and laws. Usage-based billing alone can not solve the bottleneck problem in the rush hour . A step further with UIL, a fully dynamic tariff is proposed. This dynamic price is displayed on the user interface (screen) when required (before use and billing) so that the user can decide for himself whether he wants to use the service or not. The core idea is clear that a user will generate (or trigger) less traffic as the price increases. As a result, the tariff method will also change usage behavior and traffic as a whole, similar to what is to be achieved with electricity tariffs in the smart grid . In addition, UIL is even more effective than in the previous smart grid: The immediate feedback and latency in the order of seconds allows an immediate response and also a training ( conditioning ) of the user, which is aimed at better use in the future affects.

Use

The application of user-in-the-loop is possible in all fields where limited resources are consumed and where a negative impact of the use on society or the environment must be avoided. For example with excessive consumption of energy and fossil fuels .

In mobile communications there is a problem with the expected exponentially growing data rates over the next 10 years. Smartphones and notebook computers will continue to record traffic increases of 100% per year in the future. This trend has been observed since before 2010. The traditional approach of oversizing the capacity to be able to carry all traffic is becoming increasingly difficult and expensive, because future technologies ( Next Generation Mobile Networks , 4G, 5G) can never meet demand at the same rate of increase. The energy consumption and environmental objectives are also becoming increasingly important in the future. Whatever increase in capacity a new technology will offer, it will be quickly devoured by the ever faster growing demand for traffic. New technological approaches such as micro cells or femtocells will only consume more power and money. The UIL approach is orthogonal (independent) to it and does not require extensive investment or additional electrical power. UIL can also improve the spectral efficiency of a radio cell by a significant amount.

Incentives

The interface between the UIL controller and the user block consists of information and incentives . The information can simply mean that the user knows that his actions are beneficial (for the system, the community, the society). However, in most cases an additional incentive might be necessary to really get the user to change their normal behavior temporarily, because altruism is not far-reaching enough and people tend to selfish strategies in free societies (see game theory ). The dilemma is called the Tragedy of the Commons. It is therefore only reasonable to assume that the Homo oeconomicus (model) is driven in a first approximation by a maximization of the utility and that Homo cooperativus only applies to secondary effects.

Incentives can be financial aspects (e.g. a cheap tariff) or other useful bonuses that are either monetary or not. An example are the bonus miles of a frequent flyer program or payback points for each (uncomfortable) action taken by the user.

Negative incentives are also possible in the form of sanctions (e.g. additional fees), but psychology reveals that positive incentives work better.

In mobile radio networks, an incentive can be to grant the cooperating user a higher bit rate (instead of a proportional share). A sanction could come into play if the use of the system is bad for the overall goal, namely at the current time ( rush hour ) or at the current location (traffic jam, poor radio connection). The sanction has the effect that a sufficient number of users are deterred under these circumstances. Instead, the user can use the service in a better place or at a better time without such penalties .

Examples of user-in-the-loop applications

Ecological aspects

In general, UIL allows regulation towards a goal that is more ecological than if the user were acting in an uncontrolled manner. That goal can be energy consumption , consumption of fuel , food, water etc., or even soft goals like better social behavior. It's like changing the rules (the payoff) in a game theory scenario to make participants more cooperative.

The ecological aspect in the radio network is as follows. The power consumption of wireless infrastructure ( base stations , control centers) currently accounts for around 0.5% of global electrical energy demand and is therefore proportional to the CO 2 emissions. Calculated with current values, this results in an ecological footprint of 34 g CO 2 (or 17 dm³) for 1 MB of data transmitted by radio. This is called the "green index of wireless cellular communications". One bit thus corresponds to approx. 5.8 * 10 16 CO 2 molecules as a specific bit emission. Radio networks consume 0.5% of the worldwide electrical energy requirement of approx. 20 PWh in 2010. The average monthly radio traffic is 240 * 10> 15 bytes, i.e. 2880 PB in 2010. Then the energy per byte can be calculated as 0.0347 * 10 −6 kWh and that is 0.125 J. If the electricity is obtained with coal, then 975 g CO 2 are released for 1 kWh of energy. This results in 0.0338325 mg CO 2 for one byte , which, when extrapolated, results in 34 g CO 2 for 1 MB or 34 kg CO 2 for 1 GB.

See also

Web links

Individual evidence

  1. ↑ a b c Rainer Schoenen, Halim Yanikomeroglu: User-in-the-Loop: Spatial and Temporal Demand Shaping for Sustainable Wireless Networks. In: IEEE Communications Magazine. February 2014.
  2. ↑ a b c Rainer Schoenen, Halim Yanikomeroglu, Bernhard Walke: User-in-the-Loop: Mobility Aware Users Substantially Boost Spectral Efficiency of Cellular OFDMA Systems. In: IEEE Communications Letters. volume 15, number 5, 2011, ISSN  1089-7798 , pp. 488-490.
  3. ↑ a b c d Rainer Schoenen, Gurhan Bulu, Amir Mirtaheri, Halim Yanikomeroglu: Green Communications by Demand Shaping and User-in-the-Loop Tariff-based Control. In: Proceedings of the 2011 IEEE Online Green Communications Conference (IEEE GreenCom'11). 2011, ISBN 978-1-4244-9519-1 .
  4. ↑ a b c Rainer Schoenen: On increasing the spectral efficiency more than 100% by user-in-the-control-loop. In: Proceedings of the 16th Asia-Pacific Conference on Communications (APCC). October 2010.
  5. ↑ UMTS Forum Report 44. Mobile traffic forecasts 2010–2020. http://www.umts-forum.org/
  6. ↑ Cisco Systems Inc., Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2010–2015. February 1, 2011.
  7. ↑ Sandvine Inc., 2010 Mobile Internet Phenomena Report. White paper. 2010.
  8. ↑ Rysavy Inc., Mobile Broadband Capacity Constraints And the Need for Optimization. White paper. February 2010.
  9. ↑ M. Dohler, RW Heath, A. Lozano, CB Papadias, RA Valenzuela: Is the PHY layer dead? In: IEEE Communications Magazine. April 2011, volume 49, number 4, pp. 159-165.