My initial comment on this is that I agree with you that net income cannot be constant even if the Gross income is fixed. With age, maintenence costs will vary from year to year and this will affect the net income. Also, a lease will typically have a known duration if it is for a term-of-years in which case, 1/i is probably not the approach. Unless the property market is a dead one where the forces of demand and supply are non-exisitent, then the income can be constant but no such market exisits.

1/i where 100/R  = i in valuation is an expression of the value today obtained from a simple arithmetic expression of recent sales prices divided by the net income of similar properties CV/NI = R. If future growth potentials are added, what then is the value of the investment to the purchaser. The purchaser has an interest in the income growth potentials of the property which is what the seller is terminating by his decision to sell today, so his income can only remain constatnt. If on the other hand, the valuation is for compensation due to loss through acquisition, then the future value potentials add to his loss and can be capitalized at an assumed realistic lifespan of the building expressed in years and allowing for depreciation over time..

I think that I might understand your question. Let me try to restate it: How can it be that valuation methods that assume income is constant in perpetuity be relevant? The answer is simple. These sorts of valuation methods were developed in an environment in which computations were much more difficult so appraisers tried to find closed forms of equations that simplified computations. Of course, computers allow us to inject much more complexity today. Cash flows can take on various patterns and we can compute the results very quickly. We can allow risky magnitudes through the use of monte carlo analysis. 

If you want to use the old methodologies that have rigid assumptions such as constant and never-ending income, then try to think in real terms. That is, think in terms of inflation-adjusted magnitudes. At the same time, you must use real rates.

There are a number of strands to this issue.  First it could be a logical approach if we lived in a world of very long leases with fixed rents on net terms.  If you lived in the UK in the first half of the 20th century that was the world we lived in - we had very long 99, 63, 42 year leases with no rent reviews, no expectation of growth, etc.  In those days the approach was a defensible cash flow approach.  But now of course we don't live in those times so the reason this approach survives is because it has ceased to become a logical investment based technique, it has turned into a comparison based technique where the unit of comparison is basically an initial yield.  That yield is not a rate of return in any logical sense, it is obtained from the relationship between the current rent (and/or rental value) and the sale price of the asset.  If you have no or few transactions, it is very difficult to determine this yield hence the movement towards cash flows and the determination of target rates of return.  But where you have a deeply traded market of basically similar assets, determining the exchange price (Market value) by reference to similar sales using the cap rate comparison approach has a logic to it.  But it does not identify true underlying Investment Value, only the exchange price. 

We can identify a logical initial yield from basic investment criteria; i.e. target rate - expected growth rate = initial return.  But they are the same inputs we need for a cash flow so we may as well do the job properly and set out all the cash flows so we can build in all the other assumptions that make it more logical such as potential voids/vacancies, capital expenditure, costs, etc.

one has to find out the logical input if it's not available. i think, one has to find out the rate of return on property on historic data of market value and net income . 

Neither real nor realistic. This is one of the issues that the growth-explicit valuation models seek to address.