Dynamic resource planning and allocation in a satellite communication system

Abstract

Dynamic planning and allocation of satellite resources Method of communication by satellite with equipment allocated to users, a territory covered for the purposes of the communication method being composed of sub-zones defined by satellite means and each associated with distinct allocated radio resources, within limits of total available resources planned in advance by a planning (100) of the resources. The communication method comprises a step of modification (210, 250), by a resource manager (200), of an allocation to the sub-zones of the resources, in response to the reception of a message (T1, T2) from a mobility manager (300) of the users signaling a modification of communication conditions for the users in at least one sub-zone. Abstract figure: 3

Description

Title of the invention: Planning and dynamic allocation of resources in a satellite communication system Technical background

[0001] The invention falls within the framework of two-way communications between ground stations using a satellite relay. Such communications use a transmission system which is desired to be secure, so that the exchanges cannot be listened to by third parties. These communications are carried out with modem networks (a modem being defined as a device combining the modulation and demodulation functions) to which a planning capacity is added. Modem networks must guarantee quality of service in all planned situations, including in cases of strong accidental or deliberate interference. Guaranteeing access to the service is essential in an operational context, for example military.

[0002] The context of the invention is that of ground coverage by several satellite beams, which is commonly called “multi-spot” communication, a spot being the ground footprint of a beam from a given satellite. In addition, ground stations move (it is commonly said that they are in a mobile situation) and can therefore leave the footprint of one beam and / or enter the footprint of another beam.

[0003] Conventionally, network management involves the allocation of resources which are frequency bands, radio transmission powers and hardware resources, consisting in particular of countable modems and which are qualified as radio resources or satellite resources.

[0004] It has been proposed in the prior art to allocate power according to the weather, the real-time position of the users (nevertheless assumed to be fixed) and knowledge of the instantaneous channel.

[0005] It has also been proposed in the prior art to perform a dynamic allocation of resources to improve the fairness of the flow rates between users by minimizing the difference between traffic demand and capacity.

[0006] It has also been proposed in the prior art to size resources to be allocated per spot by considering the worst case, i.e. by predicting the maximum number of users for each spot. The planning reserves the resources by allocating the resources of each spot in a fixed and independent manner. This is not suitable for scaling up, i.e. increasing the number of spots and terminals. Satellite resources are over-reserved, which is costly in cases of dense ground station deployments.

[0007] US2020 / 0328804_Al (also published under number US11664888_B2) discloses a distribution of frequency bands in a multi-spot system, according to user needs using a centralized manager. Definition of the invention and associated advantages

[0008] Known solutions provide for sizing of resources per spot by considering the most unfavorable case, namely in particular the maximum number of users per spot in all the spots, and simultaneously a maximum bit rate and a maximum symbol rate, these constraints being taken into account independently of each other, but this is not suitable for multi-spot coverage, since the resources are then over-reserved, which is very costly in cases of dense deployment.

[0009] The solutions considered impose high computational costs and are not transposable to scale.

[0010] Thus, we lack a solution that allows us to redistribute resources from one spot to another in real time while guaranteeing quality of service.

[0011] To solve the problems mentioned above, a method of communication by satellite is then proposed with equipment allocated to users, a territory covered for the purposes of the communication method being composed of sub-zones each defined by satellite means (in the sense of means on board a satellite, generally a satellite antenna directed towards said territory, transmitting on several beams, each beam forming a sub-zone on the ground) and each associated with distinct allocated radio resources, within the limits of total available resources defined beforehand by resource planning.

[0012] The communication method is remarkable because it comprises a step of modification, by a resource manager, of an allocation to the sub-zones of distinct resources, in reaction to the reception of a message from a user mobility manager signaling a modification of communication conditions for one or more users in at least one sub-zone, the resource planning having previously defined said limits of the total resources available in a logic of pooling resources between the sub-zones compatible with different allocations of distinct resources.

[0013] This method provides efficiency and resource savings while ensuring the quality of services in a multi-spot context. It provides a dynamic and agile implementation to adapt to the mobility of stations between spots. It allows optimization of the allocation of powers and spectrum, according to a criterion that can be adapted according to operational needs, such as for example the reduction of a total transmitted power.

[0014] According to advantageous and optional characteristics,

[0015] - said resources may comprise separate demodulation means as well as a separate power and a separate spectral band for satellite transmission and reception, and are within the limits of available demodulation means, available power and available spectral band.

[0016] - the user mobility manager in the territory can receive subzone manager messages reporting, for a subzone manager associated with a given subzone, for example a. a movement of at least one user out of the given sub-area or an entry of a user into the given sub-area, or b. a change in a requested quality of service in the given sub-area, or signaling a change in an atmospheric condition or a phenomenon of jamming or interference affecting radio transmissions in the given sub-area, or c. a user-initiated change of communication frequency band in the given sub-area.

[0017] - the modification step may comprise a calculation based on an objective function allocation aimed at maximizing transmission capacity, maximizing energy efficiency, fairness between sub-zones, or saving material resources, on the scale of all or part of said sub-zones.

[0018] - resource planning can be carried out upstream of a commitment of the communications according to said method.

[0019] - the resource manager can establish a new allocation by resolving a single or multidimensional bin-packing problem formulated within each sub-zone, in view of said message from the mobility manager.

[0020] The invention also consists of a satellite communication system comprising equipment allocated to users, a territory covered by the communication system being composed of sub-zones each defined by satellite means and each associated with distinct allocated radio resources, within the limits of total resources available.

[0021] Remarkably, the communication system is characterized in that it comprises a resource manager configured to allocate resources to the sub-zones, in response to the reception of a message from a mobility manager, included in the communication system, configured to signal to the resource manager a change in communication conditions for the users in at least one sub-zone.

[0022] - optionally, but advantageously, the communication system can include a dynamic allocation calculator within the resource manager and a scheduling calculator upstream of the resource manager which implement heuristic calculation methods to determine an allocation of radio resources to said sub-zones.

[0023] To ensure improved operation of the network, the invention proposes that the allocation to each spot of radio resources including hardware means (number of modems), transmission powers and the spectral band is done in such a way as to save resources while guaranteeing a quality of service required for all communications in the mobility zone. This involves improving resource planning by determining upstream the minimum radio resources necessary to guarantee performance with a given level of quality of service, by taking into account a possibility of sharing resources across all the spots and by planning to exploit such pooling of these resources between the spots.This also involves dynamic allocation (in real time) of inter-spot resources, taking into account in particular the needs of stations varying over time (due in particular to the mobility of stations moving from one spot to another, or moving within a spot) and environmental conditions such as the presence of interference and / or adverse weather conditions.

[0024] The invention makes it possible to optimize the planning and allocation or reallocation in real time of these resources, while guaranteeing access to the services. It proposes a global solution combining multi-spot planning which provides for a pooling of resources on the spots, with a solution for the dynamic allocation and distribution of these resources on the spots. These resources are reallocated as soon as there is a change in the spatial distribution of traffic on the spots. The solution thus makes it possible to save the necessary resources.

[0025] The solution is resource-optimized, more robust against jamming and interference, and adapted to the changing needs over time related to the mobility of stations. The solution is dynamic and agile. Communications are also protected during mobility. List of figures

[0026] [Fig.l] shows a resource allocation according to the prior art.

[0027] [Fig.2] shows a resource allocation according to an implementation of a mode of carrying out the invention.

[0028] [Fig.3] presents an architecture for implementing the principles of the invention.

[0029] [Fig.4] shows a sequence of implementation of an embodiment of the invention. Detailed description

[0030] [Fig.l] In [Fig.l], a configuration according to the prior art is represented, which implies, as we will see, an over-reservation of satellite resources. The allocation of resources is in fact static.

[0031] A satellite 10 is in orbit above an area of ​​the planet, which includes the mobility zone of the communication stations of different users of a communication network.

[0032] The mobility zone is covered by several spots 1, 2, 3 (the number of spots may be higher), and other spots may also be present. The term spot designates an area covered by a beam emitted by a satellite, here satellite 10, or more precisely by the satellite's antenna(s). The spots may overlap.

[0033] The communication network comprises a control station 15 which communicates with the satellite 10, and which has a control and management role. The control station 15 may be outside the spots.

[0034] User stations 20 are deployed on the ground or near the ground - and in any case not in orbit like the satellite 10. Some user stations 20 move so as to pass for example from one spot 1, 2, 3 to another, sometimes being in an area common to two spots, and sometimes leaving a spot in which it was initially present to join another spot. Thus, the stations are not uniformly distributed over the spots and their spatial distribution over the spots is variable over time, including rapidly.

[0035] In the figure there are respectively 2, 1 and 5 user stations 20 in spots 1, 2 and 3. They are represented in the form of land military vehicles, but they can be pedestrians, aircraft or boats.

[0036] The control station 15 is, like the user stations 20, deployed on the ground or near the ground - and in any case not in orbit like the satellite 10.

[0037] Network controllers 40 (in English: network controllers or NC) implemented locally within the control station 15 manage the scheduling operations and manage the resources, each for a different spot, within the framework of a virtual local area network satVLAN associated with the spot concerned.

[0038] Each spot is orchestrated by a network controller 40 distinct from the control station 15. A network controller 40 comprises a first modem 41 and possibly other modems 42 (the first modem and each of the other modems constitute distinct demodulation capacities), which serve to increase the demodulation capacities of the station.

[0039] When a station 20 is in a particular spot, and except in cases of overlap, it cannot communicate with the network controllers 40 which control the other spots in the control station 15. Thus, it can only be controlled by the network controller 40 of the spot in which it is located, and can transmit data to this network controller 40 only.

[0040] As shown in the form of arrows on the left side of the diagram, the frequency band f and the satellite power P (in terms of equivalent isotropic radiated power PIRE) allocated to each spot are identical fractions of, respectively, the total band and the total satellite power, available for the entire satellite network. The entire band and all the power are generally allocated, of course. Here, these fractions are therefore one third, since three spots are defined.

[0041] Another modem 42 has been provided for each spot, which constitutes a dimensioning ready to manage the situation in which all the user stations 20 (here the 8 user stations 20) would be present simultaneously in the same spot, without any reallocation of these other modems being provided. There are three other modems 42 here, one for spot 1, the other for spot 2, and the third for spot 3, each comprising several demodulators. The number of spots and the number of modems are nevertheless presented solely for illustration, without their being intended to define the invention.

[0042] Ultimately, due to the static nature of resource allocation, there is an over-reservation of these resources.

[0043] [Fig.2] In [Fig.2], a dynamic allocation of satellite resources is shown.

[0044] At a given time, where there are as previously respectively 2, 1 and 5 user stations 20 in spots 1, 2 and 3, another modem 42 is allocated to the network controller 40 of spot 3 in the control station 15, since it is spot 3 which covers the largest number of user stations 20. And none are allocated (any) to the network controllers 40 of spots 1 and 2 because they do not need them, their first modems 41 being sufficient to take charge of the stations covered by these spots at the time concerned, which are in limited numbers, since only 2 and 1 respectively.

[0045] Still dynamically, the frequency band f allocated to each spot is a variable fraction and different from one spot to another of the total band available on the scale of the satellite network and the power P (in terms of equivalent isotropic radiated power) allocated to each spot is a variable fraction and different from one spot to another of the available power. The entire band and all the power are generally allocated, as previously. As shown in the form of arrows in the central part of the diagram, here, the fractions are the largest for spot 3, the smallest for spot 2, and intermediate between those of these two spots 2 and 3 for spot 1, since it is spot 3 which covers the largest number of stations, and spot 2 which covers the smallest number.

[0046] [Fig.3] In [Fig.3], a new architecture for allocating resources between multiple spots is presented. It allows multi-spot planning which, based on dynamic resource allocation, wastes less resources than existing solutions, and dynamic resource allocation in real time which did not exist in this context before.

[0047] This architecture allows the steps of, initially, planning carried out by a planning calculator 100, then in a second step of dynamic allocation of resources, as well as triggering (“triggers”, which are signaling messages) which lead to signaling exchanges between the deployed stations (the user stations 20 of [Fig.2]) and an orchestrator 200 responsible for orchestrating the resources of the networks.

[0048] Planning aims to provide a maximum budget of resources to be allocated upstream of deployment. By resources, we mean transmission powers, spectral bands and material resources.

[0049] The planning is implemented in the planning calculator 100 which takes as input static configuration elements 110 comprising the data linked to the static configuration such as the topology of the spots, the quality of service requirements per station, the multidimensional transmission constraints (bit rates, symbol rates, etc.), the number of stations deployed and the access methods used (e.g. Time Division Multiple Access, TDMA). Other parameters can be taken into account for the planning.

[0050] This resource planning aims to determine the minimum radio resources necessary to guarantee performance with a level of quality of service, and to take into account the fact that the resources must be shared across the different spots, whereas currently there is an over-reservation of resources spot by spot.

[0051] One of the objectives pursued by the planning, which is expressed in the form of a planning objective function (or objective function) 120 is to improve the planning, that is to say to minimize the resources to ensure a quality of service required for all the communications in the mobility zone.

[0052] Other optimization objectives or constraints are considered in certain variants. These objectives expressed in the form of the planning objective function (or objective function) 120 may aim for better resistance to interference in certain geographical areas, or even favor the optimization of other criteria such as capacity, energy efficiency, fairness between spots, battery economy. Other objectives or constraints are envisaged.

[0053] Constraints can be defined differently depending on the type of resources, depending on the spots and the geographical areas, such as for example to take into account the existence and location of geographical areas of particular sensitivity to interference.

[0054] The planning objective may integrate particular hypotheses of user mobility zones between spots according to the mobility profile of the stations. Thus, the stations may be known as not leaving a spot, not very mobile from one spot to another, or on the contrary very mobile between spots.

[0055] The optimization objective can also introduce a distinction according to the geographical area of ​​the stations in the spots. For example, according to a probability of presence in the spot.

[0056] The planning calculator 100 thus defines a budget of allocated resources 150.

[0057] The orchestrator 200 receives the static configuration elements 110, and also information per spot in real time such as radio data linked to the propagation channel (weather, interference, jammer, etc.). It also receives trigger messages.

[0058] A trigger is a message indicating a significant modification at the level of the deployed stations. Following receipt of a trigger by the orchestrator 200, the resources (all the resources or a subset of resources defined by a type of resources, such as for example the frequency bands, the hardware, the allocated power) are distributed to the spots jointly to all the spots using a dynamic resource allocation calculator 210 which takes into account the change indicated to the orchestrator 200 in the trigger.

[0059] The dynamic allocation calculator 210 initially received a budget of allocated resources 150 defined by the planning calculator 100.

[0060] The dynamic allocation calculator 210 distributes, i.e. dynamically allocates (in real time) the radio resources by taking into account the needs which vary over time (due in particular to the mobility of the deployed stations), and the environmental conditions such as the presence of jamming and / or the weather.

[0061] The triggers are sent to the orchestrator 200 by a mobility manager (MObility Manager or MOM in English) 300. Generally, the orchestrator 200 and the mobility manager 300 have signaling exchanges.

[0062] The mobility manager 300 communicates directly with the network controllers 40 (see [Fig.2]) or 40_l, 40_12, ... 40_N which provide it with real-time information linked to their spot, by messages M.

[0063] Two examples of triggers are described.

[0064] Trigger 1 Tl is sent by the mobility manager 300 to the orchestrator 200 when there is a change in the spatial distribution of the stations in the spots.

[0065] Trigger 2 T2 is sent by the mobility manager 300 to the orchestrator 200 when the capacity proves insufficient on a spot. This may be linked to a change quality of service or bit rate, or a change in radio conditions, for example.

[0066] This information, spot by spot, reaches the mobility manager 300 in the form of messages M and the mobility manager 300 synthesizes it, and generates a trigger T1 or T2 if and when this is appropriate, in view of an overall understanding of the radio transmission conditions in the spots, as obtained by the aggregation of successive messages M (including from a single message M, if the content thereof is sufficient to justify the transmission of a trigger T1 or T2).

[0067] The passages from one frequency band to another are equivalent to an intercellular transfer (Handover in English). They request the mobility manager 300 which receives an M message. For example, if a station changes from a Ka band configuration to the X band, the mobility manager 300 is requested by the reception of an M message. Indeed, changing frequency band, for a station, causes the station to leave a modem subnetwork so that it can connect to another modem subnetwork. It is therefore useful to transfer radio resources from the left subnetwork to the joined subnetwork.

[0068] As for the planning carried out by the planning calculator 100, the resource distribution policy carried out by the dynamic allocation calculator 210 can also be pre-configured, for example by the satellite network operator, and aim, using an objective allocation function 220, at maximizing capacity, maximizing energy efficiency, fairness between spots, saving batteries, or other objectives.

[0069] We take into account the pooling of all resources (material, power and spectrum) between all satellite spots or sub-networks.

[0070] In a variant, a Trigger 3 T3 is sent to the orchestrator 200 by a network controller 40 and contains real-time information relating to the spot concerned specifically such as local meteorological elements, the local propagation of radio waves around the spot or even specific quality of service (QoS) requirements and updated at the date and time and instant concerned, at the spot.

[0071] Once the resources have been allocated by the dynamic allocation calculator 210, the orchestrator 200 transmits the decided allocations to each network controller 40 by a resource allocation message 250.

[0072] [Fig.4] The architecture thus proposed is implemented. [Fig.4] presents the sequential sequence during the deployment of this architecture.

[0073] During the planning phase: on the basis of (static) information elements 110 on the configuration or the topology, the multi-spot planning module 100 calculates the resource budget allocated to all the spots and sends this information 150 to the real-time resource orchestrator.

[0074] During the operational phase of allocation or re-allocation of resources: the resource orchestrator 200 dynamically distributes this resource budget to the different network controllers 40_i, by the dynamic allocation 250 transmitted by the dynamic allocation calculator 210.

[0075] This operation is performed on the basis of information Ta, Tb, Te provided by the mobility manager (MObility Manager) 300, in particular the distribution of stations on the spots, and from radio information T3 provided in real time by each spot controller, in particular radio conditions, weather, etc.

[0076] The reallocation of resources between spots is triggered by 2 triggers notifying a change in spatial distribution or even insufficient capacity on a spot and which are here the information Tb and Te. Principles implemented

[0077] In one embodiment, only the hardware resource is optimized (i.e. the number of modems, and more precisely the number of other modems 42, (see [Fig.2]). The other resources to be allocated, in particular the power and the frequency band are fixed and dimensioned independently in each spot. Still in this embodiment, each modem has maximum reception flow constraints. No constraint is imposed on the bit rate, but only on the symbol rate (but the opposite could be done, or constraints imposed simultaneously on these two rates). In addition, each network controller 40 (see [Fig.2]) includes the main modem (which is unique) and any number, which may be 0, of other modems, the symbol rates of the stations are independent of the spot in which the stations are located at each instant, and the transmission by a station to which radio resources are allocated is done continuously (no TDMA time division access in this embodiment). Each modem has the role, in particular, of demodulating the carriers transmitted by the stations deployed in the spots.

[0078] It is desired to allocate the carriers received by the network controller 40 to hardware means using as few modems as possible.

[0079] The constraints related to modems are the fact that each modem has a maximum capacity in symbol rate and / or bit rate, and the fact that a modem can only demodulate a maximum number of carriers, i.e. can only support a limited number of stations.

[0080] During the planning stage, the number of stations to be deployed is known, but not their distribution in the spots since it varies over time after the initial deployment.

[0081] An existing solution considers the worst-case distribution, i.e. a maximum number of users per spot. The existing algorithm then calculates for each spot the number of modems required for this worst-case distribution.

[0082] However, we want to determine in each spot the distribution of stations in the modems using the fewest modems possible. Solving this problem is equivalent to solving a bin-packing problem, a variant of the knapsack problem, which is a combinatorial problem. It is proposed here to use heuristics, genetic algorithms, or reinforcement learning methods to give an approximate solution.

[0083] In an example of 8 stations distributed over 3 spots, the number of modems required is calculated by considering 8 stations in a spot. If this number is 3 for example, since there are 3 spots, the existing solution would schedule 9 modems in total. Since this single-spot scheduling method wastes a large quantity of modems, a new multi-spot scheduling solution is proposed here.

[0084] Details of the planning calculator 100 and the dynamic allocation calculator 210 in an exemplary embodiment

[0085] Planning

[0086] The planning carried out by the planning calculator 100 takes into account the pooling of all the resources between all the spots (or satellite sub-networks) thanks to a resource distribution solution described here, to minimize the resources while ensuring a quality of service required for all the communications in the mobility zone.

[0087] Other optimization objectives or constraints at the planning stage are taken into account in different variants, such as resistance to interference in certain geographical areas, or a choice to optimize the use of certain resources to the detriment of the use of others.

[0088] The optimization objective at the planning stage may impose specific constraints on each type of resource (or even by spot or geographic area), such as the presence of an area of ​​particular sensitivity to interference.

[0089] This optimization objective at the planning stage can integrate specific hypotheses relating to user mobility zones between the spots according to the mobility profile of the stations (fixed in a spot, little or very mobile between the spots). The optimization objective can introduce a distinction according to the geographical zone of the users in the spots.

[0090] A heuristic implemented in one embodiment for scheduling involves optimized scheduling of the number of modems by oversizing symbol rates instead of the number of stations. This results in reduced modem overbooking, and reduced complexity. This also results in scalability through oversizing symbol rates.

[0096] The rating

[0091] A heuristic is thus proposed that exploits multi-spot resource sharing. Symbol rates are therefore oversized rather than the number of stations.

[0092] The heuristic contains two steps: a step of generating a reduced table of the possibilities of distributing the spots and a step of calculating the number of modems required for these possibilities.

[0093] Let nUE be the number of stations to be deployed and M the number of spots.

[0094] The heuristic takes into account all possible distributions of stations in the spots made up of stations and reduces the number of possible cases by considering the most unfavorable case in terms of symbol rate.

[0095] As an example, we consider nÜE = 8 stations and M = 3 spots. The 8 station flow rates are in the example, in decreasing order 5 MBd, 4 MBd, 4 MBd, 3 MBd, 2 MBd, 1 MBd, 1 MBd and 1 MBd. \ means that there are respectively Pf P2' P3 stations distributed 1' *2' <3 / in spots 1, 2, 3.

[0097] We consider the distribution (6, f ]). There are 6 stations in the first spot, 1 in the second and one in the third, which corresponds to the announced total of 8 stations.

[0098] As each station can have a different associated symbol rate, the distribution (6.1, 1) therefore encompasses a large number of possibilities for distributing the stations in the spots. To reduce this number of possibilities, one solution consists of over-sizing the symbol rates by considering the following notation: [6ma» lma® Imax]' oij P”1** with P natural integer designating the P stations with the highest symbol rates among the nUE stations deployed. The distribution [6max- Imax, Imax] s'Sn*9u we consider the 6 stations with the highest symbol rates in a first spot. These 6 highest rates are 5 MBd, 4 MBd, 4 MBd, 3 MBd, 2 MBd and 1 MBd in the example.

[0099] The station with the highest throughput, in this case 5 MBd, is also placed in spot 2. It is also placed in spot 3. It is therefore artificially placed three times.

[0100] The two stations with the lowest flow rates are not placed in any spot. Here they are two stations with a flow rate of 1 MBd.

[0101] This approach can be applied to spots 1, 2, 3 in this order, but also 1, 3, 2, or 2, 1, 3, ...

[0102] In the following, the notation [6max. lmax, lmaJ indicates the following distribution: the 6 stations with the highest symbol rates in a spot A, the station with the highest symbol rate also in spot B and also in spot C.

[0103] By using the described approach, the number of distribution possibilities to be considered is greatly reduced.

[0104] In the following table, we list a part of all the distribution possibilities in spots A, B, C:

[0105] [Tab.l] Table 1 presents an example of distribution of the planning heuristic according to an embodiment for 8 stations and 3 spots. Distribution No. Spot A Spot B SpotC 1 ^max ^max 2 7 1 max ^max 3 A 'hnox? —max 4 ^max l / nox 5 ^max -max Qwax 6 ^max ? ^max lmax 7 4 ^max 4 ^max Umax

[0106] The first phase of the heuristic therefore consists of generating this table. This can be done with, in one embodiment, an iterative algorithm. It is possible to generate this table for nUE = 100 stations and M = 7 spots in 2 minutes.

[0107] The second phase of the heuristic consists of calculating the number of modems for the distributions of stations in the spots indicated in Table 1.

[0108] The idea is to apply a bin-packing resolution method in the given configurations. This can be done with a first-fit decreasing (FFD), next-fit, or best-fit algorithm. The FFD heuristic is advantageous because of its speed of execution.

[0109] Subsequently, we note BPP(nmax) the number of modems required calculated with the method of solving the bin-packing problem for the stations with the highest speeds.

[0110] The number of planned modems MEp]aiq for a distribution i provided in the generated table can therefore be calculated as follows: [ 0111 the ME |)1 " u = L^ I BPP(n„,„„), where BPP(0 m ")=l

[0112] In fact, at least one modem is required per spot and nm.max corresponds to the number of stations considered (those with the highest speeds) in the spotm.

[0113] For the distribution [g, il, the planned number of modems is therefore: L îïldX? BlaX» ITiclXj [0H4] MEpian, = BPP(6max) + BPP(lmax) +BPP(lf„J

[0115] The number of modems planned is then determined by taking the most unfavorable case among the possible distributions indicated by the generated table, namely:

[0116] MEp[an = maxMEp|anj

[0117] The planned number of modems ME_max is always greater than the total number of modems required in real time, for all possible distributions of stations in the spots. This statement can be proven mathematically.

[0118] Another solution for planning

[0119] A heuristic implemented in another embodiment, still for planning, uses reinforcement learning. This is proposed here because the problem addressed has been identified as a combinatorial optimization problem and modeled with a Markovian decision process.

[0120] Dynamic allocation

[0121] In response to the dynamic allocation calculator receiving a trigger, resources, which may be resources of all resource types or a subset of the resources, are allocated or reallocated to the spots.

[0122] In the case of trigger T2 (see [Fig.3]), there is capacity on a spot which turns out to be insufficient, due to a change in the required quality of service or a change in the bit rate.

[0123] It is necessary to reallocate a number of modems adapted to the guarantee of the required quality of service. In the example considered, it is assumed that the transmission power and the frequency bands are fixed and oversized, and that the constraints specific to the modems considered are known: namely a maximum number of carriers and fixed maximum symbol rates.

[0124] For each spot, the problem of distributing stations in modems is solved by one of the heuristics next fit, best fit, or first fit.

[0125] In one variant, the distribution is solved by reinforcement learning applied to 3D bin packing, taking into account the dimension of the bit rate, symbols and other parameters.

[0126] In the case of trigger T1 (see [Fig.3]), there is a change in the spatial distribution of the stations in the spots, which is notified to the dynamic allocation calculator by the mobility manager. This is therefore a change in the real-time distribution of the stations in the spots.

[0127] A dynamic resource allocation solution is chosen by solving a bin-packing problem within each spot. The bin-packing problem is solved using the FFD heuristic, or another heuristic with low execution time.

[0128] In a situation for n_UE=8 stations and M=3 spots and where in real time, stations 1, 2 and 3 are in spot 1 with bit rates (symbol rates) of 4, 3 and 5 MBb respectively, stations 4, 5 and 6 are in spot 2 with bit rates of 1, 2 and 4 MBb respectively and stations 7 and 8 are in spot 3, with bit rates of 1 and 1 MBb respectively, it is chosen to allocate another modem 42 to the network controller 40_l, and none to the network controllers 40_2 and 40_3.

[0129] A trigger variant Tl may be transmitted due to handover information from one frequency band to another, in the case of multi-band spots. Such a trigger is also notified to the dynamic allocation calculator by the mobility manager.

[0130] In the case of trigger T3, there is for example a change in radio conditions.

[0131] Here too, for each spot, the problem of the distribution of stations in the modems is solved by heuristics or reinforcement learning.

[0132] The invention can be generalized by considering several characteristics in addition to or instead of the symbol rate and by solving a multidimensional bin-packing problem. General comments

[0133] Generally speaking, it is specified that the optimization of the allocation of resources also takes into account, in certain embodiments, the access methods used by the stations, such as for example TDMA (time division multiple access technique).

[0134] Furthermore, the present presentation has discussed demodulation means, but it must be borne in mind that these are accompanied by reception means, and also by modulation and transmission means.

Claims

Claims

1. Method of communication by satellite with equipment (20) allocated to users, a territory covered for the purposes of the communication method being composed of sub-zones (1, 2, 3 ...) defined by means on board a satellite (10) and each associated respectively with distinct allocated radio resources, within limits of total available resources defined beforehand by a resource planning, the communication method being characterized in that it comprises a step of modification (210, 250), by a resource manager (200), of an allocation to the sub-zones (1, 2, 3, ...) of distinct resources, in reaction to the reception of a message (T1, T2) from a mobility manager (300) of the users signaling a modification of communication conditions in at least one of said sub-zones (1, 2, 3, ...), resource planning having previously defined the said limits of the total resources available in a logic of pooling resources between the sub-zones (1, 2, 3, ...) compatible with different allocations of distinct resources.

2. A method of satellite communication according to claim 1, characterized in that said resources comprise separate demodulation means (40) as well as separate power (P) and separate spectral band (f) for satellite transmission and reception, and are within limits of available demodulation means, available power and available spectral band.

3. A method of satellite communication according to claim 1 or claim 2, characterized in that the mobility manager (300) of the users in the territory receives messages (M) from sub-area managers (40; 40_i) signaling, for a sub-area manager (40; 40_i) associated with a given sub-area (1, 2, 3, ...), a movement of at least one user outside the given sub-area (1, 2, 3, ...) or an entry of a user into the given sub-area (1, 2, 3, ...), or signaling a modification of a quality of service requested in the given sub-area (1, 2, 3, ...), or signaling a modification of an atmospheric condition or of a jamming or interference phenomenon affecting the radio transmissions in the given sub-area (1, 2, 3, ...), or signaling a change of communication frequency band initiated by a user in the given sub-area (1, 2, 3, ...).

4. Method of communication by satellite according to one of claims 1 to 3, characterized in that the modification step (210, 250) comprises a calculation (210) based on an objective allocation function (220) aiming at maximizing transmission capacity, maximizing energy efficiency, fairness between sub-zones, or saving material resources, on the scale of all or part of said sub-zones.

5. Method of communication by satellite according to one of claims 1 to 4, characterized in that the resource manager (200) establishes a new allocation by solving a mono or multidimensional bin-packing problem formulated within each sub-zone (1, 2, 3, ...), in view of said message (T1, T2) from the mobility manager (300).

6. Satellite communication system comprising equipment allocated (20) to users, a territory covered by the communication system being composed of sub-zones (1, 2, 3, ...) defined by means on board a satellite (10) of the system and each associated with distinct allocated radio resources, within limits of total available resources previously defined by a resource planning calculator (100), the communication system being characterized in that it comprises a resource manager (200) configured to allocate resources to the sub-zones (1, 2, 3, ...), in response to the reception of a message (T1, T2) from a mobility manager (300), included in the communication system, configured to signal to the resource manager (200) a modification of communication conditions for the users in at least one of said sub-zones (1, 2, 3, ...), the planning calculator (100) having previously defined said limits of the total resources available in a logic of pooling resources between the sub-zones (1, 2, 3, ...) compatible with different allocations of distinct resources.

7. Satellite communication system according to claim 6, characterized in that it comprises a dynamic allocation calculator (210) within the resource manager and a planning calculator (100) upstream of the manager. of resources (200) which implement heuristic calculation methods to determine an allocation of radio resources to said sub-zones.

Images