Context : The Kepler Object of Interest Network ( KOINet ) is a multi-site network of telescopes around the globe organised for follow-up observations of transiting planet candidate Kepler objects of interest ( KOIs ) with large transit timing variations ( TTVs ) .
The main goal of KOINet is the completion of their TTV curves as the Kepler telescope stopped observing the original Kepler field in 2013 .

Aims : We ensure a comprehensive characterisation of the investigated systems by analysing Kepler data combined with new ground-based transit data using a photodynamical model .
This method is applied to the Kepler-82 system leading to its first dynamic analysis .

Methods : In order to provide a coherent description of all observations simultaneously , we combine the numerical integration of the gravitational dynamics of a system over the time span of observations with a transit light curve model .
To explore the model parameter space , this photodynamical model is coupled with a Markov chain Monte Carlo algorithm .

Results : The Kepler-82b /c system shows sinusoidal TTVs due to their near 2:1 resonance dynamical interaction .
An additional chopping effect in the TTVs of Kepler-82c hints to a further planet near the 3:2 or 3:1 resonance .
We photodynamically analysed Kepler long- and short-cadence data and three new transit observations obtained by KOINet between 2014 and 2018 .
Our result reveals a non-transiting outer planet with a mass of m _ { f } = 20.9 \pm 1.0 M _ { \Earth } near the 3:2 resonance to the outermost known planet , Kepler-82c .
Furthermore , we determined the densities of planets b and c to the significantly more precise values \rho _ { b } = 0.98 _ { -0.14 } ^ { +0.10 } \text { g cm } ^ { -3 } and \rho _ { c } = 0.494 _ { -0.077 } ^ { +0.066 } \text { g cm } ^ { -3 } .

Conclusions :