Patent Application
20150201348
The invention relates to the field of wireless telecommunications and networks more particularly the management of a cellular, mesh and ad hoc telecommunications network.
More and more frequent usage of terminals connected to a cellular network has caused an upsurge in traffic.
To counteract this rise in traffic, in the case of a cellular network, operators have boosted the density of cells in a given geographic region to offer a certain quality of service (QoS) to users.
A cell is defined by a base station which offers radio coverage for terminals inside the cell as defined.
It is specified that the term cell designates equally an attocell, a femtocell, a pico-cell, a micro-cell, or even a macro-cell.
One problem is that in a given geographic area the densification of cells has given rise to an increase in energy consumption of networks, with some areas of the network being over-equipped.
Solutions for defining spatio-temporally managing the network in a given geographic area are known.
Reference could be made to the document by Wei, Y., Song, M., Liu, B., Wang, X., & Li, Y. (2011, October): “Energy-efficient cooperative relaying and cognitive radio technologies to deliver green communication”, in Pervasive Computing and Applications (ICPCA), 2011 6th International Conference on (pp. 105-109), IEEE.
One solution is to manage the network as a function of the day or of the night based on the principle that the demand in traffic will be less at night than by day: the power of base stations can be reduced for example at night.
This approach is described in the following document: Chiaraviglio, L, Ciullo, D., Meo, M., & Marsan, M. A. (2009, September): “Energy-efficient management of UMTS access networks”, in Teletraffic Congress, 2009, ITC 21 2009. 21st International (pp. 1-8), IEEE.
Other solutions are based on statistical information obtained via the traffic request passed to a given geographic region over a given time period. In this way, there are request profiles defined over and space, which manage the network.
But these solutions are unsatisfactory, as they do not take into consideration the traffic request in a given region in real time. So these solutions often result in or over-dimension or under-dimension the network, with under-dimensioning leading to degradation of the QoS.
Also, the problem of the taking the evolution of the traffic request into consideration in real time in a given region turns up again in wireless and ad-hoc mesh networks.
Consequently there is a need to be able to configure a mesh or ad hoc cellular network to limit energy costs and at the same time guarantee quality of service for users.
An aim of the invention is to manage a network as anticipated.
For this purpose, the invention proposes a method for configuration of a telecommunications network located in a geographic area, which has at least one terminal receiving or transmitting traffic relative to a service, the method comprising:
The invention is advantageously completed by the following characteristics taken singly or in any technically possible combination:
The invention is based on a prediction of traffic from information which when the management of the network is based on past measures are not taken into account.
In fact, precise and local estimation of the capacity requested reconfigures the network to avoid situations known as “out of service” and boost QoS. Also, reconfiguration can be done with the aim of optimising the network according to one or more parameters (such as energy efficiency) selected by the operator.
The invention applies to wireless networks: cellular (3G, GSM, GPRS, etc.), ad hoc and mesh networks.
Other characteristics, aims and advantages of the invention will emerge from the following description which is purely illustrative and non-limiting, and which must be seen in light of the appended drawings, in which:
a to 4e illustrate two embodiments of a method according to the invention;
In all figures similar elements bear identical reference numerals.
In relation to
The telecommunications network can be of mesh type and comprises at least one terminal located in a geographic area. This type of network differs from that of
Terminal means a telephony terminal a sensor, a computer, etc.
At least one terminal T1, T2, T3, T4, T5, T6, T7, T8 receives or transmits traffic directed to another terminal.
In the case of a cellular network (as illustrated in
The configuration can be set up centralised. For this, a controller is linked to all access points of the geographic area and acts as master of the network, the access points being slaves.
Alternatively, management can be carried out as distributed. In this case, the elements of the network of the geographic area communicate with each other to be configured relative to each other.
In relation to
It is specified here that traffic means packets of data (for example IP packets) transmitted or received by the terminal, the traffic corresponding to a requested or received service by the terminal.
Information relative to traffic relates to the content of the traffic and to the sending/receipt context of the traffic.
It is considered that traffic can originate from various applications (or services): telephony, video, SMS, video game.
More generally, these can be all types of services which need resources from the telecommunications network.
The context of sending/receipt of traffic can be the location of the terminal in the geographic area, the quality of the radio link, the type of communication (outside or inside what has a degree of mobility of the terminal in the geographic area), the base stations covering the mobile terminal in the geographic area (for example the terminal can be in an area covered by several base stations, in the case of a cellular network), attenuation of the radio signal, level of interference, etc.
The content of the traffic relates to the type of data coming from the services. The traffic is not identical if it is a telephone or a video call.
It is noted that the context is already utilised in various applications described in the documents below:
The content and context can consequently let the network determine potential chaining of different types of content. This estimation is already implemented by the services of type “web streaming video”, (for example youtube®) and music (for example deezer®). These services propose certain types of content of video/music type which have a rapport with the preceding user requests. Also, the network can begin transmitting data relative to these predictions to reduce wait time and therefore improve the quality of the service. This case comes up especially for reading lists set up by a user.
The context information used here is information linking a terminal to its current use context, especially information stating that the terminal is inside or outside, or future, for example obtained from the history of the terminal, that is, information drawn from experience: for example regular shifts by the terminal, at particular time slots.
According to an embodiment, traffic sent or received by each terminal comprises information relative to the traffic.
According to this embodiment, advantageously the acquired information S1 comprises a duration of the requested or received service, a duration relative to the content.
With respect to the duration of the service, it can be of two orders: if the service requested is a reading service or continuous diffusion (“streaming”) wanted instantaneously (therefore which cannot be deferred), the duration of the service corresponds to the real duration of the information to be obtained by reading; if the service requested is likely to be deferred, the duration of the service can correspond to the quantity of data to be exchanged to provide this service.
Advantageously, this information S0 is inserted into the traffic directly on sending of the traffic. In this way, the information relative to traffic is transmitted explicitly by the terminal. In the case of a cellular network, information relative to the traffic can be inserted with signalling data. In this latter case the communication standard has to be adapted.
Alternatively or in addition, each terminal comprises an application loaded in memory of the terminal to retrieve information relative to traffic.
So, the method for configuration comprises an extraction step S0′ of information relative to traffic, and a transmission step S0″ of information relative to traffic. Such an application can be installed by the operators themselves before selling a terminal or can be provided to users so they install it on their terminal. In this last case a counterpart can be proposed to users (reduction on their subscription, for example). Therefore, it is the application which explicitly sends S0″ the information relative to traffic.
According to another embodiment, ensuring acquisition S1 data relative to traffic are extracted on receipt of traffic as such by a controller of the network, by a deep packet inspection (DPI) technique. Such a technique analyses traffic to establish statistics, to detect intrusions from spam or any other type of content. Such a technique is classically used for the Internet and is now applied to telecommunications networks. In this way, reference could be made to the document by R. Bendrath, “Global technology trends and national regulation: Explaining Variation in the Governance of Deep Packet Inspection,” International Studies Annual Convention, New York City, 15-18 Feb. 2009.
On completion of the acquisition step S1, a criterion characteristic of the possibility of deferring said requested or received service over time is determined at S2, from acquired information.
This criterion depends on the type of service according to whether this is real-time traffic or which can wait (for example large-volume email).
Next, a traffic request profile arriving in the geographic area over a time period after acquisition is estimated S3 from acquired information and the criterion as to the possibility or not of deferring the service.
The request profile is the capacity of the network requested in a given geographic area over a given period.
For this, the estimation step S3 consists of updating an initial request profile from information relative to the traffic acquired. This initial request profile is a function of the profile of each user.
The profile of each user comes in particular from past observations which take habits into account.
In other terms, the initial request profile P0 is the average traffic profile which characterises the given geographic region as a function of users present in the given geographic area.
Next, the new traffic request profile can be represented as a source of traffic to be served by the cellular network by queuing techniques and in particular technique called weighted fair queuing (WFQ) or else called priority queuing (PQ).
In this way, reference could be made to document: A. Parekh and R. Gallager: “A Generalized Processor-Sharing Approach to Flow Control in Integrated Services Networks: The Single-Node Case”, IEEE/ACM Transactions on Networking, Vol. 1, No 3, June 1993 and R. Rönngren and R. Ayani: “A comparative study of parallel and sequential priority queue algorithms”, ACM trans. Model. Comput. Simul. 7, 2, April 1997, pages 157-209.
The estimation S3 can be performed according to different time scales:
Of course, the shorter the estimation duration, the greater the precision.
Therefore, the traffic request profile is a prediction of traffic to arrive at the geographic area and, contrary to known techniques which are based on past observations, the information relative to the content and to the context of incoming traffic is also used here.
As illustrated in this figure the profile is three-dimensional: time, space capacity.
Once the request profile is obtained, a configuration of the network which produces an offer profile closest to the traffic request profile is determined S4.
This determination can be done under restriction. In this case, the offer profile closest to the request profile which best satisfies the restriction to be fixed will be determined.
Processes for automatic learning to determine the configuration of the network which best responds to the request profile can be used. In this respect, reference could be made to document WO 2007/057857 A1 (KONINKL PHILIPS ELECTRONICS NV [NL]. PHILIPS CORP [US]. RIBAS SALVADOR) (2007 May 24).
Configuration of the network is defined by all base stations activated and by their characteristics (number of antennas, power, calculation and cooperation capacity between base stations, etc.).
According to an embodiment, the restriction can be that the energy consumption of the configuration of the network ensuring a given level of QoS is minimal. In this way, the restriction on the offer profile ensures the required quality of service and the restriction on configuration ensures minimal energy consumption.
The restriction on the configuration of the network for having the minimal energy consumption can be defined either by a function of cost or by a function of benefit or again by a function of gain.
The function of cost can be defined by CF=Σiwi*(Pi/Pmax) where Pi is at least one of the following parameters: capacity, timeframe, wait, packet loss rate, transition time, transitory energy, Pmax is the maximal value of Pi, wi is the weight associated with each parameter Pi. Other metrics relate to the packet error rate, the quantity of energy per bit (expressed in Joules per bit), etc.
The benefit metric can be defined by BF=Σiwi*(Ei) where Ei is the energy efficiency (bit/joule) associated with each component of the network, wi is the weight associated with each Ei. Other metrics relate to spectral efficiency (expressed in bit/s/Hz), deployment efficiency (expressed in bits/euros (or dollars)), etc.
Alternatively or in addition, the restriction can be temporal or spatial. In other terms, for a certain duration and/or as a function of a region of the geographic area, one or the other of the restrictions hereinabove can be selected.
The different optimisation restrictions can be fixed by the operator in charge of deployment of the network.
Once the configuration of the network enabling an offer profile closest to the request profile is determined, S5, the network is configured.
Configuration of the network consists of adapting the number of active base stations, adapting the sending power of base stations, etc.
In the precise case of a mesh network also configuration of defining the itinerary relative to each traffic flow (defined by the routing function).
In other terms, a method for configuration of a telecommunications network located in a geographic area, comprising the following steps, is provided here:
An extra step of physical configuration of the network according to the configuration as determined can then be provided.
Prior to the physical configuration step, an optimisation step of the estimated theoretical profile can be provided, by taking into account the use context of the terminal and the content (possibility of deferring the service and its duration).
An optimisation step of the configuration of the network can also be provided as a function of predetermined criteria, for example the consumption of the network.
To clarify these different steps, several concrete examples will now be described within the scope of wireless communications (ideas coming from these examples apply also to other types of telecommunications networks described earlier), by supposing that in a telecommunications network a wireless communications terminal requires a service.
In a first example, this service is instantaneous streaming type, that is, it involves a need for instantaneous transmission and over a determined period, for example reading a video. This service cannot be deferred, but additional content information can be identified: the duration of the video in question. As a function of the position of the terminal at the time of request of the service (context), base stations will be activated to respond to this service in the environment of the terminal, and over the duration of the video. An increase in the capacity of the network can therefore be anticipated during the reading time of the video, then a drop in the capacity of the network at the end of the video period.
In a second example, the relevant service is known as being likely to be deferred (content information). There is for example the case of downloading a set of data of predetermined size (duration of the service). In this case, it can be deduced from context information that a number of configurations of the network in the future can be used to provide the service. The information drawn from experience of the terminal can also be used to increase the scope of these possibilities. In particular, if it is known that the terminal travels a similar path every day between different types of base stations more or less saturated over time allocated to provide the service, several possible configurations of the network can determined over time to provide the service. The choice of configuration finally used can be made at random, or by optimisation of predetermined criteria such as consumption of the network, as described hereinbelow.
Two embodiments of the method according to the invention are described in relation to
In the following:
In relation to
The traffic request profile in terms of capacity at the instant t0 is sketched by the vertical bars rising above the terminals UE1, UE2, UE3 in
The information relative to the traffic (context and content) acquired and the initial request profile estimate a traffic request profile in the relevant geographic area at an instant t1 (short term).
The information relative to the traffic received or sent (information relative to context and content) deduced from information relative to traffic listed above estimate a traffic request profile in the relevant geographic area at an instant t1 (short-term).
This traffic request profile at the instant t1 is sketched in
As seen, it is evident that the terminal UE4 will request a certain capacity.
To determine the configuration of the network for defining an offer profile closest to the traffic request profile, there are three solutions possible:
To determine the optimal configuration network, determining the offer profile of traffic is done under the restriction that the configuration has minimal energy consumption for the requested level of QoS.
For this, the energy consumption of each element of the network is determined. In particular, the consumption Pin of base stations AP1, AP2 and M-BS is determined as follows: Pin=P0+ÜPTX with P0 the power of the base station at minimal load, Ü the dependence on power consumption per load and the necessary power PTX radio frequency at the base station to satisfy the request profile.
It is noted that P0 and Ü depend on the type of base station. In particular, it is very small for the local base stations (AP1, AP2 and AP3) and mush larger for the base station of macrocell M-BS. In fact, the energy consumption of a local base station depends minimally on its load and transmission power PTX. By comparison, the energy consumption of a base station of macrocell M-BS is quasi proportional to its transmission power PTX (or to its load).
In the case above, the solution according to which the transmission power of local the base station AP1 is increased is the most efficient solution from the energy point of view of the network and at the same time ensures the quality of service requested for the terminal UE4. In fact, activating the local base station AP2 or increasing the load of the base station of macrocell M-BS would visibly increase the energy consumption of the network. Also, activation of the local base station AP2 needs greater reaction time than the latency time available to serve the terminal UE4.
c illustrates the configuration of the resulting network. It is evident relative to
As a variant, from acquired information at the instant t0, the information relative to the acquired traffic (context and content) and the initial request profile estimate a traffic request profile in the relevant geographic area at an instant t2 (average-term).
This traffic request profile at the instant t2 is sketched in
With the difference of the traffic request profile estimated for the instant t1:
To determine the configuration of the network for defining an offer profile closest to the traffic request profile, two solutions are possible:
To determine the optimal configuration network, determining the traffic offer profile is done on condition that configuration of the network ensuring a given level of quality of service for users has minimal energy consumption.
Under this condition, the solution consisting of activating the local base station AP2 and placing the local base station AP1 on standby after linking of the terminal UE4 to the local base station AP2 (handover procedure) produces the network configuration closest to the request profile and has minimal energy consumption.
e illustrates the configuration of the resulting network. It is evident relative to
An example which illustrates performances in terms of energy consumption of a network configured as per the method of the invention will now be described.
A network of femtocells deployed in a building which can be represented by a 5×5 grid, twenty cellular users are located in the grid and require high-rate video traffic will be considered.
The femtocells are capable of detecting the presence/absence of a user and being deactivated from/placed on standby.
The active femtocells are characterised by energy consumption Pin=P0+ÜPTX with P0,Ü and PTX which indicate respectively the power of the femtocell at minimal load, the dependence of the power consumption per load and the power radio frequency needed to satisfy the request profile.
By comparison, the femtocells on standby have no data to transmit (no traffic) and are characterised by energy consumption Pidle<Pin.
The knowledge of the content of the traffic is exploited to allow the femtocells to deactivate only when a user a of traffic is to be transmitted and put on standby when the associated user has no more traffic to receive.
P0=4.8 W, Pidle=2.9 W and Ü=15.
PTX depends on the number of frequency resources used by the femtocell (NR) according to: PTX=NR*PRF, with PRF which is equal to 100 mW.
The energy consumption is calculated as a function of the parameter ρd which measures the probability that a femtocell is installed in an apartment.
These simulations show that a gain of 50% is obtained when the configuration of the network is determined according to the invention.
An example of use of the method of the invention in the case of a mesh network deployed to offer a high-speed wireless Internet connection to passengers of high-speed trains will now be described.
Terminals UE1, UE2 are located in a train moving along a track alongside which base stations AP1, AP2, AP3, AP4 are deployed.
According to a first step, information on the context of user terminals such as their position in the train, the course of the train, the speed of the train and the content of the traffic generated by the applications used by these terminals is acquired.
This information is then aggregated per user and per geographic area then used to estimate one traffic request profile per user then per area in a later given time period.
A benefit function is then defined from the energy consumption of each base station as a function of the number of user terminals served by each base station and the type of traffic expected.
Finally, a configuration of the network for attaining a given minimal level of QoS and maximising said benefit function is determined. In this scenario, the information on the context (that is, the position, course and speed) can be exploited to:
These two functions reduce energy consumption, squandering of resources, latency and prevent failures in the handover process which in real time changes the association between terminals and base stations.
Also, the information on traffic content can be exploited to set up/allocate in real time the capacity (that is, the bandwidth) required/necessary in a specific part of the network.
This approach increases the efficiency of the system and therefore the number of users which can be served by the network.
| Number | Date | Country | Kind |
|---|---|---|---|
| 1256933 | Jul 2012 | FR | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2013/065230 | 7/18/2013 | WO | 00 |
